U.S. patent application number 13/061398 was filed with the patent office on 2011-06-30 for silicate glass article with a modified surface.
This patent application is currently assigned to AALBORG UNIVERSITET. Invention is credited to Morten Mattrup Smedskjaer, Yuanzheng Yue.
Application Number | 20110159219 13/061398 |
Document ID | / |
Family ID | 40510556 |
Filed Date | 2011-06-30 |
United States Patent
Application |
20110159219 |
Kind Code |
A1 |
Yue; Yuanzheng ; et
al. |
June 30, 2011 |
SILICATE GLASS ARTICLE WITH A MODIFIED SURFACE
Abstract
The present invention relates to a silicate glass article, such
as a glass container, with a modified surface region. The modified
surface has, among other advantageous properties, an improved
chemical durability, an increased hardness, and/or an increased
thermal stability, such as thermal shock resistance. In particular
the present invention relates to a process for modifying a surface
region of a silicate glass article by heat-treatment at T.sub.g in
a reducing gas atmosphere such as H.sub.2/N.sub.2 (1/99). The
concentration of network-modifying cations (NMC) in the surface
region of the silicate glass article is lower than in the bulk
part, and the composition in the surface region of the
network-modifying cations is a consequence of an inward
diffusion.
Inventors: |
Yue; Yuanzheng; (Aalborg,
DK) ; Smedskjaer; Morten Mattrup; (Aalborg,
DK) |
Assignee: |
AALBORG UNIVERSITET
Aalborg O
DK
|
Family ID: |
40510556 |
Appl. No.: |
13/061398 |
Filed: |
September 3, 2009 |
PCT Filed: |
September 3, 2009 |
PCT NO: |
PCT/DK2009/050224 |
371 Date: |
February 28, 2011 |
Current U.S.
Class: |
428/34.4 ;
501/35; 501/38; 501/53; 501/54; 501/64; 501/68; 501/72;
65/30.14 |
Current CPC
Class: |
C03C 23/007 20130101;
C03C 2218/35 20130101; Y10T 428/131 20150115 |
Class at
Publication: |
428/34.4 ;
65/30.14; 501/53; 501/54; 501/64; 501/68; 501/72; 501/35;
501/38 |
International
Class: |
C03C 21/00 20060101
C03C021/00; B32B 1/02 20060101 B32B001/02; C03C 3/04 20060101
C03C003/04; C03C 3/06 20060101 C03C003/06; C03C 3/095 20060101
C03C003/095; C03C 3/083 20060101 C03C003/083; C03C 3/078 20060101
C03C003/078; C03C 13/00 20060101 C03C013/00; C03C 13/02 20060101
C03C013/02 |
Foreign Application Data
Date |
Code |
Application Number |
Sep 5, 2008 |
DK |
PA 2008 01249 |
Claims
1. A silicate glass article comprising a bulk part, a surface
region, and network-modifying cations (NMC): wherein the silicate
Mass article has a weight percentage of polyvalent metal oxides of
0.5-30%; wherein the silicate glass article comprises a polyvalent
element selected from the group consisting of: Au.sup.3+,
Au.sup.2+, Au.sup.+, Ir.sup.3+, Pt.sup.2+, Pd.sup.2+, Ni.sup.2+,
Rh.sup.+, Rh.sup.3+, Co.sup.2+, Co.sup.3+, Mn.sup.4+, Mn.sup.3+,
Ag.sup.3+, Ag.sup.2+, Ag.sup.+, Se.sup.6+, Se.sup.4+, Se,
Ce.sup.4+, Cr.sup.6+, Cr.sup.4+, Cr.sup.3+, Cr.sup.2+, Sb.sup.5+,
Sb.sup.3+, Cu.sup.3+, Cu.sup.2+, Cu.sup.+, U.sup.4+, Fe.sup.6+,
Fe.sup.3+, Fe.sup.2+, As.sup.5+, As.sup.3+, As, Te.sup.7+,
Te.sup.4+, Te, V.sup.5+, V.sup.4+, V.sup.3+, Bi.sup.4+, Bi.sup.3+,
Bi.sup.2+, Bi.sup.+, Eu.sup.3+, Ti.sup.4+, Ti.sup.3+, Sn.sup.4+,
Sn.sup.2+, Zn.sup.2+, and Cd.sup.2+; wherein the concentration of
the network-modifying cations in the surface region is lower than
in the bulk part; wherein the silicate bridging-oxygen content is
higher in the surface region than in the bulk region; and wherein
the composition in the surface region of the network-modifying
cations is a consequence of an inward diffusion.
2-28. (canceled)
29. The silicate glass article according to claim 1, wherein the
silicate glass article has a weight percentage of silica of at
least 50%.
30. The silicate glass article according to claim 1, wherein the
silicate glass comprises transition metallic cations.
31. The silicate glass article according to claim 30, wherein at
least some of the transition metallic cations are network-modifying
cations (NMC).
32. The silicate glass article according to claim 30, wherein the
transition metallic cations are selected from a group consisting
of: Ti.sup.4+, Ti.sup.3+, V.sup.5+, V.sup.4+, V.sup.3+, Cr.sup.6+,
Cr.sup.5+, Cr.sup.3+, Mn.sup.7+, Mn.sup.6+, Mn.sup.5+, Mn.sup.4+,
Mn.sup.3+, Fe.sup.5+, Fe.sup.4+, Fe.sup.3+, Co.sup.4+, Co.sup.3+
and Ni.sup.3+.
33. The silicate glass article according to claim 30, wherein the
transition metallic cations are selected from a group consisting
of: Ti.sup.2+, V.sup.2+, Cr.sup.2+, Mn.sup.2+, Fe.sup.2+,
Co.sup.2+, Ni.sup.2+, Cu.sup.2+, Zn.sup.2+, Zr.sup.2+, Nb.sup.2+,
Mo.sup.2+, Ru.sup.2+, Rh.sup.2+, Pd.sup.2+, Ag.sup.2+, Cd.sup.2+,
Ta.sup.2+, W.sup.2+, Re.sup.2+, Os.sup.2+, Ir.sup.2+, Pt.sup.2+,
Hg.sup.2+ and Ra.sup.2+.
34. The silicate glass article according to claim 1, wherein at
least some of the network-modifying cations (NMC) are from Group
IIa in the Periodic Table.
35. The silicate glass article according to claim 1, wherein said
silicate glass article is a glass container, a glass fiber, art
glass, or a glass container capable of storing a liquid.
36. A process for modifying a surface region of a silicate glass
article, said process comprises the step of heat-treating the
silicate glass article in an atmosphere comprising a reducing gas,
wherein the silicate glass article has a weight percentage of
polyvalent metal oxides of 0.5-30%, wherein the silicate glass
article comprises a polyvalent element selected from the group
consisting of: Au.sup.3+, Au.sup.2+, Au.sup.+, Ir.sup.3+,
Pt.sup.2+, Pd.sup.2+, Ni.sup.2+, Rh.sup.+, Rh.sup.3+, Co.sup.2+,
Co.sup.3+.sub., Mn.sup.4+, Mn.sup.3+, Ag.sup.3+, Ag.sup.2+,
Ag.sup.+, Se.sup.6+, Se.sup.4+, Se, Ce.sup.4+, Cr.sup.6+,
Cr.sup.4+, Cr.sup.3+, Cr.sup.2+, Sb.sup.5+, Sb.sup.3+, Cu.sup.3+,
Cu.sup.2+, Cu.sup.+, U.sup.4+, Fe.sup.6+, Fe.sup.3+, Fe.sup.2+,
As.sup.5+, As.sup.3+, As, Te.sup.7+, Te.sup.4+, Te, V.sup.5+,
V.sup.4+, V.sup.3+, Bi.sup.4+, Bi.sup.3+, Bi.sup.2+, Bi.sup.+,
Eu.sup.3+, Ti.sup.4+, Ti.sup.3+, Sn.sup.4+, Sn.sup.2+, Zn.sup.2+,
and Cd.sup.2+, wherein the heat-treatment is performed at 0.7-2.0
times the glass transition temperature (T.sub.g) of the silicate
glass, said process resulting in an inward diffusion of the
network-modifying cations (NMC) into deeper regions of the silicate
glass article, whereby the concentration of the network-modifying
cations in the surface region is lowered, said process resulting in
the formation of a silicate bridging-oxygen content that is
substantially higher in the surface region than in the bulk
region.
37. The process according to claim 36 wherein the reducing gas is a
mixture of reducing gasses.
38. The process according to claim 36, wherein the reducing gas is
further mixed with one or more inert gasses.
39. The process according to claim 36, wherein the atmosphere
comprises a mixture of nitrogen gas and hydrogen gas.
40. The process according to claim 36, wherein the atmosphere
comprises a mixture of carbon monoxide gas and carbon dioxide
gas.
41. The process according to claim 36, wherein the atmosphere
comprises a mixture of gasses selected from a group consisting of:
SbH.sub.3, AsH.sub.3, B.sub.2H.sub.6, CH.sub.4, PH.sub.3,
SeH.sub.2, SiH.sub.4, SH.sub.2, SnH.sub.4, Cl.sub.2, NO, N.sub.2O,
CO, H.sub.2, N.sub.2O.sub.4, SO.sub.2, C.sub.2H.sub.4, and
NH.sub.3.
42. The process according to claim 36, wherein the heat-treatment
is performed so as to obtain a thickness of said surface region of
at least 100 nm.
Description
TECHNICAL FIELD OF THE INVENTION
[0001] The present invention relates to a silicate glass article,
such as a glass container, with a modified surface region. The
modified surface has, among other advantageous properties, an
improved chemical durability, an increased hardness, and/or an
increased thermal stability, such as thermal shock resistance. In
particular, the present invention relates to a process for
modifying a surface region of a silicate glass article.
BACKGROUND OF THE INVENTION
[0002] It is well known that surface characteristics have strong
impact on the physical and chemical properties of glasses, and
hence, on their applications. These properties can be tailor-made
by using a surface modification technique, e.g., coating of metal
oxides or polymers, ion exchange between glass and salt melt, fire
polishing and so on. By modification of the surface, new functional
surfaces can be created that can be applied in new contexts or in
marked improvements of existing materials.
[0003] Earlier studies have shown that the surface of a
Fe.sup.2+-bearing silicate glass article can be modified by using
redox reactions. The new surface can be obtained by heat-treating
the said glass article in atmospheric air at temperatures near the
glass transition temperature (T.sub.g) for suitable durations. The
heat-treatment leads to oxidation of ferrous iron (Fe.sup.2+) to
ferric iron (Fe.sup.3+), which causes a diffusion of divalent
cations (primarily Mg.sup.2+) from the interior of the glass
towards the surface (called outward diffusion). Surprisingly, the
oxidation process does not cause oxygen to diffuse into said glass
article to a significant degree as Fe.sup.3+ is formed due to
reaction between electronic species (electron holes) and Fe.sup.2+.
A crystalline layer forms on the surface of the glass article as
the divalent cations react with oxygen at the surface. This surface
layer exhibits excellent thermal performance, i.e., this finding
has potential to be industrially applied. The effect of the surface
layer on the physical and chemical properties (mechanical
properties, chemical durability, inertness, optical properties,
etc.) of glassy materials is still unknown.
[0004] Another study has shown that the physical and chemical
properties of a glass article can be affected by the surface
content of silica, as silica increases the connectivity of the
glass structure [Deriano et al., 2004]. For example, an enhancement
of the chemical durability of the glass is expected. Hence, their
surface modification method has potential to be industrially
applied, e.g., in the improvement of the chemical resistance of
glass container for drugs and chemicals. Glass degradation can
occur when liquids are corrosive.
[0005] In yet another study, Pind & Sorensen (2004) studied the
redox and diffusion processes in a
SiO.sub.2--Al.sub.2O.sub.3--MgO--CaO--FeO/Fe.sub.2O.sub.3 glass
system and prepared samples with Fe.sup.3+/Fe.sub.tot ratios of
approximately 80%. One of these samples was heated in a reducing
atmosphere (10/90 H.sub.2/N.sub.2). In this case, Mg.sup.2+
diffused from the surface towards the interior (called inward
diffusion) as a mirror-image of the oxidation mechanism. The
silicon concentration near the surface is increased as the silicon
ions (Si.sup.4+) do not diffuse, i.e., a silica-rich nanolayer
forms on the surface. Other studies dealing with reduction of
ferric iron-bearing glasses in H.sub.2 atmospheres have shown that
permeation (dissolution and diffusion) of gaseous H.sub.2 is the
dominant reduction process, i.e., no inward diffusion of divalent
cations occurs [Gaillard et al., 2003].
[0006] In a study by Rigato et al. (1994), it was shown that the
thermochemical reduction of alkali-lead-silicate glass does not
lead to any significant incorporation of hydrogen in the surface,
but greatly sensitizes the surface to the chemical and physical
adsorption of water. The treatment creates a thin (25 nm)
compositionally modified layer, perhaps microporous, of silica-rich
glass at the surface due to depletion of Pb where the hydrogen
concentration due to adsorption is irreversible. The time and
temperature of thermochemical treatment influence the initial
kinetics of the adsorption. These observations are of practical
significance to the behaviour of electron multiplier and
microchannel plate devices that have been exposed to humid
environments.
[0007] In a study by Deriano et al. (2004), it is described that
mechanical properties of a soda-lime-silica glass can be improved
by thermal treatment of the glass in N.sub.2 and NH.sub.3 gases.
They argue that the observed strength improvement might be due to
the reaction of water and ammonia with the glass. This causes an
exchange process of network-modifying cations with protons. Hereby,
the soda-lime-silica glass is turned into a glass of a high silica
content by a process that is rate-limited by the diffusion of
water.
[0008] The U.S. Pat. No. 3,460,927 describes a thermal treatment
method that improves the flexural strength (the ability to resist
deformation under load) of a polyvalent element-containing glass by
reducing it in a hydrogen atmosphere. The treatment is carried out
well below the glass transition temperature.
[0009] The understanding of the relation between redox reactions
and diffusion processes near the surface of iron-containing
silicate glasses in a reducing atmosphere is at present very
limited due to results pointing in different directions.
[0010] If the surface of high silica content can be created, this
would be economically more favourable than bulk silica (SiO.sub.2)
production as the latter requires very high temperatures (up to
2400.degree. C.) for melting and forming.
[0011] Hence, an improved method to create silicate glass products
with a relatively thick surface of high silica content would be
advantageous as an economically more favourable option than using
the expensive bulk silica in the production, or as a favourable
option to the use of coatings of metal oxides or polymers, ion
exchange between glass and salt melt, fire polishing and so on.
SUMMARY OF THE INVENTION
[0012] Thus, an object of the present invention relates to
improving the properties of silicate glass articles.
[0013] In particular, it is an object of the present invention to
provide an improved silicate glass article that solves the above
mentioned problems of the prior art with improved surface
properties.
[0014] Thus, one aspect of the invention relates to a silicate
glass article comprising a bulk part and a surface region, said
silicate glass article comprises network-modifying cations (NMC);
wherein the concentration of the network-modifying cations in the
surface region is lower than in the bulk part, wherein the
composition in the surface region of the network-modifying cations
is a consequence of an inward diffusion.
[0015] The invention is particularly, but not exclusively,
advantageous for obtaining an improved silicate glass article
having improved chemical durability, an increased hardness, and/or
an increased thermal stability. Without being bound by any specific
theory, it is contemplated that the network modifying cations (NMC)
occupy interstitial positions within the network and thereby create
non-bridging oxygens. By lowering the concentration of the network
modifying cations (NMC) in the surface region, a more connected
network is created on the surface, which makes it difficult for
ions to diffuse through the glass, and thus improves the chemical
durability, e.g. acid and alkali resistance.
[0016] Similarly, the mechanical properties, e.g. the hardness, are
augmented due to the increased connectivity of the surface layer
resulting in an increased effective silica concentration in the
said surface layer.
[0017] It could be an advantage, in addition to the increased
connectivity of the surface region, to increase the thickness of
said surface region to further improve the chemical durability, to
increase the hardness, and/or to increase the thermal stability. To
obtain a relatively high concentration of silica in the surface
region, it may be contemplated that the glass type comprises a
relatively large weight percentage of silica such as 10%, 20%, 30%,
40%, 50%, 60%, 70%, 80%, or 90%.
[0018] Therefore, in one embodiment, the silicate glass article
according to the invention has a weight percentage of silica of at
least 10-35%, preferably at least 30-49%, and even more preferably
at least 50%. Other components than silicate may be comprised in
the silicate glass article, such as alkali oxides, alkaline earth
oxides and polyvalent metal oxides. In another embodiment, the
silicate glass article according to the invention has a weight
percentage of alkali oxides of at least 0-90%, such as 0.5-85%,
preferably at least 1-80%, such as 3-75%, preferably at least
5-50%, such as 7-30%, preferably at least 10-20%. In yet another
embodiment, the silicate glass article according to the invention
has a weight percentage of alkaline earth oxides of at least 0-90%,
such as 0.5-85%, preferably at least 1-80%, such as 3-75%,
preferably at least 5-50%, such as 7-30%, preferably at least
10-20%.
[0019] In still another embodiment, the silicate glass article
according to the invention has a weight percentage of polyvalent
metal oxides of at least 0.001-90%, such as 0.5-85%, preferably at
least 1-80%, such as 3-75%, preferably at least 5-50%, such as
7-30%, preferably at least 10-20%.
[0020] The surface layer exerts a strong impact on the surface
properties as silica increases the connectivity of the glass. In
particular, it considerably enhances the chemical durability (in
both acid and alkali solutions) and the hardness of the glass.
[0021] Therefore, in another embodiment, the silicate glass article
according to the invention has a silicate bridging oxygen content
that is substantially higher in the surface region than in the bulk
region, i.e. the network connectivity of the surface region is
higher than that of the bulk region.
[0022] In particular, in one embodiment of the invention, the
silicate glass article according to the invention has a decrease in
the number of non-bridging oxygen atoms per tetrahedron, NBO/T, in
the surface region of at least 10%, 20%, 30%, 40%, 50%, 60%, 70%,
80%, 90%, or 100%.
[0023] In yet another embodiment, the silicate glass article
according to the invention has a concentration of SiO.sub.2 in the
surface region that is substantially higher than in the bulk
part.
[0024] Oxidation of an iron-bearing glass by thermal treatment in
atmospheric air causes the Mg.sup.2+, Ca.sup.2+, and Fe.sup.2+ ions
to diffuse from the interior towards the surface (called outward
diffusion). This observation is consistent with the results of
previous studies based on basaltic glass systems. The diffusion of
Mg.sup.2+ is predominant in the overall diffusion process, and at
the surface, Mg.sup.2+ ions react with external oxygen to form
periclase (MgO) crystals. The Fe.sup.2+ ions that diffuse to the
surface are oxidized to Fe.sup.3+ at the surface. The surface
region or layer enhances the hardness of the glass and protects it
from attack of an acid solution, but makes it more vulnerable
against an alkali solution.
[0025] A striking phenomenon has been observed as the outward
diffusion of divalent cations does not only occur under an
oxidizing atmosphere of heat-treatment, but also under N.sub.2,
even under a reducing atmosphere like H.sub.2/N.sub.2 (10/90 v/v)
at the ambient condition. The outward diffusion causes the
formation of an oxide nanolayer on the glass surface that possesses
morphology and concentration profile different from those of the
crystalline layer created when heating the glass in air. The extent
of the cationic diffusion depends on temperature and duration of
the heat-treatment. It has been proposed that the outward diffusion
in N.sub.2 and H.sub.2/N.sub.2 (10/90) is related to thermal
nitridation (nitrogen incorporation), i.e., the mechanism of the
outward diffusion depends on the type of gas used for the
heat-treatment. The reduction of Fe.sup.3+ to Fe.sup.2+ or V.sup.5+
to V.sup.4+ in H.sub.2/N.sub.2 (10/90) operates by permeation of
H.sub.2 into the glass. Thus, hydroxyl groups form and are
incorporated into the glass structure. The incorporation of
hydroxyl groups increases the rate of the cationic diffusion even
though the reduction of Fe.sup.3+ does not cause the diffusion in
H.sub.2/N.sub.2 (10/90). Furthermore, the created OH groups reduce
the stability of the glass against crystallization and the
mechanical properties of the glass.
[0026] Surprisingly, it has been found by the inventors of the
present invention that when the glasses are heated in
H.sub.2/N.sub.2 (1/99), both H.sub.2 permeation and outward
diffusion of electron holes contribute to the reduction of
Fe.sup.3+ to Fe.sup.2+ or V.sup.5+ to V.sup.4+. Diffusion of the
electron holes is charge-compensated by an inward diffusion of the
mobile network modifying cations (primarily Mg.sup.2+, Ca.sup.2+,
and Fe.sup.2+). Consequently, a silica-rich nanolayer forms on the
surface of the glass as the Si.sup.4+ ions do not diffuse.
Therefore, in a further embodiment according to the invention, the
inward diffusion is caused by reduction by a reducing gas and/or a
reducing liquid.
[0027] The thickness of the silica-rich layer can be controlled by
the content of the polyvalent element. Therefore, in another
embodiment according to the invention, the depth of the surface
region is a function of the inward diffusion process. In still
another embodiment according to the invention, the composition in
the surface region of the network-modifying cations is a
consequence of inward diffusion, wherein the inward diffusion is
caused by reduction of a polyvalent element.
[0028] It may be advantageous if the reduced element has lower
mobility than the earth alkaline ions in the glass network.
[0029] The thickness of the silica-rich layer can also be
controlled by tuning the temperature and duration of the heat.
Therefore, in still another embodiment according to the invention,
the depth of the surface region is a function of time, temperature,
field strength of diffusing ions, partial pressure of the reducing
gas, concentration and redox ratio of a polyvalent element, and/or
glass type. Hence, the layer thickness can be tuned according to
specific requirements.
[0030] A kinetic analysis has verified the diffusion mechanism of
the present invention as being characterized by chemical diffusion
and the diffusion coefficients for the divalent cations have been
calculated. Therefore, in yet another embodiment according to the
invention, the diffusion is characterized by chemical diffusion.
The diffusion is rate-limited by the reduction kinetics in a manner
where the diffusion is parabolic with time.
[0031] There could be different criteria for selecting a polyvalent
element in the production of a glass article.
[0032] The polyvalent element should in certain embodiments of the
invention have a redox state that is relatively easy to reduce in a
weak reducing atmosphere, e.g. in about 0.001, 0.01, 0.02, 0.03,
0.07, or 0.09 bar H.sub.2.
[0033] For some glass articles, the element and the redox state may
determine the color of the glass article depending on the glass
application, e.g. transparency of glass, specific color for a
specific application, art glass, specific UV-absorption to protect,
e.g., medicals, beer, wine, and other liquids or chemicals from
degradation.
[0034] In yet another embodiment, the present invention relates to
a silicate glass article, said silicate glass article being: a
glass container for storage of chemicals, a glass fiber, art glass,
a glass container for storage of beer, wine, and other liquids. In
particular, the present invention is advantageous for storage of
harsh or aggressive chemicals or for use in mechanically harmful
environments.
[0035] Therefore, in a further embodiment according to the
invention, the silicate glass is transparent in the optical range
of 10-1200 nm, preferably in the visible range of 380-750 nm.
[0036] In still a further embodiment of the present invention, the
silicate glass is capable of absorbing UV-light in the range of
between 400-10 nm, 400-315 nm, 315-280 nm, or 280-100 nm,
preferably in the range of between 400-100 nm.
[0037] The Vickers hardness (H.sub.v) test has been developed as a
method to measure the hardness of materials. In the present
invention, the Vickers hardness measurements reveal that the
heat-treated glasses are harder than the original glass. The
hardness increases with duration and temperature of the
heat-treatment, i.e., the hardness increases when the thickness of
the modified layer increases.
[0038] Therefore, in a preferred embodiment of the present
invention, the silicate glass article has a hardness of the
silicate glass that is substantially higher in said surface region
than in the corresponding surface region of untreated glass, e.g.
at least +10%, +20%, +30%, +40%, +50%, +100%, +200%, +300%, +1000%
higher H.sub.v.
[0039] The said nanolayer exerts a strong impact on the surface
properties as silica increases the connectivity of the glass. In
particular, it may considerably enhance the chemical durability in
both acid and alkali solutions. In acid solutions, leaching of
alkali ions from the glass is the dominant dissolution mechanism.
In one example of the present invention (FIG. 6B), the sodium
leached from the glass article into a HCl solution is decreased
more than five times as compared to the untreated glass.
[0040] Therefore, in a preferred embodiment of the present
invention, the silicate glass article has a chemical durability in
the said surface region that is substantially higher than in the
corresponding surface region of untreated glass, e.g. at least 1.1,
1.2, 1.3, 1.4, 1.5, 1.6, 1.7, 1.8, 1.9, 2, 3, 5, 10, 30, 50, 100,
1000 times better than in the corresponding surface region of
untreated glass.
[0041] The modified surface has, among other advantageous
properties, an increased thermal stability, such as thermal shock
resistance.
[0042] Therefore, in a preferred embodiment of the present
invention, the silicate glass article has a thermal shock
resistance that is substantially higher than the thermal shock
resistance of the corresponding untreated glass, e.g. at least 1.5,
2, 3, 5, 10, 30, 50, 100, 1000 times better than the thermal shock
resistance of the corresponding untreated glass.
[0043] The thickness of the silica-rich layer may be controlled by
the content and reduction of the polyvalent element.
[0044] In a preferred embodiment of the present invention, the
silicate glass article comprises transition metallic cations.
[0045] In still a preferred embodiment, the present invention
relates to a silicate glass article wherein at least some of the
transition metallic cations are network-modifying cations
(NMC).
[0046] In another embodiment, the present invention relates to a
silicate glass article, wherein at least some of the
network-modifying cations (NMC) are from Group IIa in the Periodic
Table, e.g. Be.sup.2+, Mg.sup.2+, Ca.sup.2+, Sr.sup.2+, Ba.sup.2+,
and Ra.sup.2+.
[0047] In still another embodiment, the present invention relates
to a silicate glass article, wherein the polyvalent element is
selected from a group consisting of: Au, Ir, Pt, Pd, Ni, Rh, Co,
Mn, Ag, Se, Ce, Cr, Sb, Cu, U, Fe, As, Te, V, Bi, Eu, Ti, Sn, Zn,
and Cd.
[0048] In another embodiment, the present invention relates to a
silicate glass article, wherein the transition metallic cations are
selected from a group consisting of: Ti.sup.4+, Ti.sup.3+,
V.sup.5+, V.sup.4+, V.sup.3+, Cr.sup.6+, Cr.sup.5+, Cr.sup.3+,
Mn.sup.7+, Mn.sup.6+, Mn.sup.5+, Mn.sup.4+, Mn.sup.3+, Fe.sup.5+,
Fe.sup.4+, Fe.sup.3+, Co.sup.4+, Co.sup.3+ and Ni.sup.3+.
[0049] In still another embodiment, the present invention relates
to a silicate glass article, wherein the transition metallic
cations are selected from a group consisting of: Ti.sup.2+,
V.sup.2+, Cr.sup.2+, Mn.sup.2+, Fe.sup.2+, Co.sup.2+, Ni.sup.2+,
Cu.sup.2+, Zn.sup.2+, Zr.sup.2+, Nb.sup.2+, Mo.sup.2+, Ru.sup.2+,
Rh.sup.2+, Pd.sup.2+, Ag.sup.2+, Cd.sup.2+, Ta.sup.2+, W.sup.2+,
Re.sup.2+, Os.sup.2+, Ir.sup.2+, Pt.sup.2+, Hg.sup.2+ and
Ra.sup.2+.
[0050] The process of the invention leads to a silicate glass
article with a surface of high silica content, thereby avoiding the
need to produce glass articles of the bulk silica glass. The latter
requires very high temperature (up to 2400.degree. C.) for melting
and forming. Therefore, the present invention is economically more
favorable than bulk silica glass production.
[0051] The invention creates an improved silicate glass article,
having improved chemical durability, an increased hardness, and/or
an increased thermal stability, without using the extrinsic coating
technology that requires additionally expensive raw materials.
[0052] Thus, another aspect of the invention relates to a process
for modifying a surface region of a silicate glass article, said
process comprises the step of heat-treating the silicate glass
article in an atmosphere comprising a reducing gas, said process
resulting in an inward diffusion of the network-modifying cations
(NMC) into deeper regions of the silicate glass article, whereby
the concentration of the network-modifying cations in the surface
region is lowered.
[0053] Yet another aspect of the invention relates to said process,
wherein the reducing gas is a mixture of one or more reducing
gasses.
[0054] Still another aspect of the invention relates to said
process, wherein the reducing gas is further mixed with one or more
inert gasses.
[0055] A preferred aspect of the invention relates to said process,
wherein the atmosphere comprises a mixture of nitrogen gas and
hydrogen gas.
[0056] Another preferred aspect of the invention relates to said
process, wherein the atmosphere comprises a mixture of carbon
monoxide gas and carbon dioxide gas.
[0057] Still a preferred aspect of the invention relates to said
process, wherein the atmosphere comprises a mixture of gasses
selected from a group consisting of: SbH.sub.3, AsH.sub.3,
B.sub.2H.sub.6, CH.sub.4, PH.sub.3, SeH.sub.2, SiH.sub.4, SH.sub.2,
Sn H.sub.4, Cl.sub.2, NO, N.sub.2O, CO, H.sub.2, N.sub.2O.sub.4,
SO.sub.2, C.sub.2H.sub.4, and NH.sub.3.
[0058] In order to obtain inward diffusion in a glass article,
caused by reduction of a polyvalent element, it is currently
considered essential that the permeation of the reducing gas is not
dominating said reduction.
[0059] A preferred aspect of the invention relates to said process,
wherein the reducing gas is substantially impermeable in the
untreated silicate glass.
[0060] It could be an advantage, in addition to the increased
connectivity of the surface region or layer, to increase the
thickness of said surface region to further improve the chemical
durability, to increase the hardness, and/or to increase the
thermal stability.
[0061] Thus, a preferred aspect of the invention relates to said
process, wherein the heat-treatment is performed so as to obtain a
thickness of said surface region of at least 100 nm, 200 nm, 400
nm, 500 nm, 600 nm, 700 nm, 1000 nm, 1500 nm, or 3000 nm.
[0062] The thickness of the silica-rich layer can be controlled by
tuning the temperature and duration of the heat-treatment.
[0063] Thus, a preferred aspect of the invention relates to said
process, wherein the heat-treatment is performed at e.g. 0.1-3.0,
0.5-3.0, 0.6-3.0, 0.7-3.0, 0.8-2.0, or 0.9-2.0 times the glass
transition temperature (T.sub.g) of the silicate glass.
[0064] Another aspect of the invention relates to said process,
wherein the heat-treatment is performed in the interval of
0.001-36, 0.01-36, 0.1-36, 0.1-30, 0.1-24, 0.2-36, 0.2-34, 0.2-20,
0.3-36, 0.3-25, 0.3-18, 0.4-36, 0.4-27, 0.4-12, 0.5-36, 0.5-15,
1-5, 1-4, or 1-3 hours. Even shorter or longer times are within the
teaching of the invention.
[0065] Regulating the pressure of the surrounding atmosphere in
said process has an important impact on the temperature and/or
duration of heat-treatment.
[0066] Yet another aspect of the invention relates to said process,
wherein the pressure of the said atmosphere in the interval of
0.001-20 atm., 0.001-10 atm., 0.01-10 atm., 0.01-5 atm., 0.1-5
atm., or 1-10 atm. Any of the lower limits in the said intervals
may also be minimum values.
BRIEF DESCRIPTION OF THE FIGURES
[0067] The invention will now be described in further details in
the following non-limiting examples.
[0068] FIGS. 1A-B show schematic representations of different
proposed mechanisms of surface modification, 1A shows the formation
of an MgO/CaO layer, 1B shows the formation of a silica-rich
layer,
[0069] FIG. 2A-D show schematic representations of SNMS depth
profiles of the untreated 6 wtFe glass and of the 6 wtFe glass
heated in H.sub.2/N.sub.2 (1/99) at different conditions, 2A shows
a profile of the original 6 wtFe glass, 2B shows a profile of the 6
wtFe glass heated at T.sub.g for 2 hours, 2C shows a profile of the
6 wtFe glass heated at T.sub.g for 16 hours, 2D shows a profile of
the 6 wtFe glass heated at 1.05 T.sub.g for 2 hours,
[0070] FIG. 3A-D show schematic representations of FT-IR
reflectance spectra of the untreated and heat-treated 6 wtFe glass
at different conditions, 3A shows a spectra of the 6 wtFe glass
heated at T.sub.g for different durations in H.sub.2/N.sub.2
(10/90), 3B shows a spectra of the 6 wtFe glass heated for 2 hours
at different temperatures in H.sub.2/N.sub.2 (10/90), 3C shows a
spectra of the 6 wtFe glass heated at T.sub.g for different
durations in H.sub.2/N.sub.2 (1/99), 3D shows a spectra of the 6
wtFe glass heated for 2 hours at different temperatures in
H.sub.2/N.sub.2 (1/99),
[0071] FIG. 4A shows a plot of squared thickness of the
divalent-cation-depleted region (.DELTA..xi.) versus heat-treatment
duration (t.sub.a) for the 6 wtFe glass samples in H.sub.2/N.sub.2
(1/99) at T.sub.g, FIG. 4B shows a table describing the dependence
of the difference in the Fe.sup.2+ concentration before and after
treatment (.DELTA.c(Fe.sup.2+)) on the initial iron-content of the
glass and the heat-treatment condition, 4C shows CEMS spectra of 6
wtFe glasses (untreated, heated in air at T.sub.g for 16 h, and
heated in H.sub.2/N.sub.2 (1/99) at T.sub.g for 16 h,
[0072] FIG. 5A shows a UV-VIS-NIR spectra of 0.20 mm thick 6 wtFe
glass samples heated in H.sub.2/N.sub.2 (1/99) at T.sub.g for
different durations, 5B shows the corresponding Fe.sup.2+
concentrations expressed as the dependence of the difference in the
Fe.sup.2+ concentration before and after treatment
(.DELTA.c(Fe.sup.2+)) on the heat-treatment duration (t.sub.a),
[0073] FIG. 6A shows a table with data of Vickers hardness and
water contact angle of the untreated and thermally treated 6 wtFe
glasses at different temperatures and durations, 6B shows a table
with data of chemical durability of the untreated and thermally
treated 6 wtFe glasses at different temperatures and durations,
and
[0074] FIG. 7 shows a schematic overview of the experimental
strategy and the employed analytical techniques, and
[0075] FIG. 8 shows viscosity as a function of temperature for the
SiO.sub.2--CaO--Fe.sub.2O.sub.3-A.sub.2O glasses with A=Na, K, Rb,
Cs, and
[0076] FIG. 9 shows Arrhenius plot of In k' as a function of the
reciprocal absolute temperature of the
SiO.sub.2--CaO--Fe.sub.2O.sub.3-A.sub.2O (A=Na, K, Rb, Cs) glasses
that have been heat-treated in H.sub.2/N.sub.2 (1/99) at different
temperatures for 2 h, inset shows the activation energy of calcium
diffusion E.sub.d as a function of both the ionic radius r.sub.A of
the alkali ions (open circles) and the fragility index m (closed
circles), and
[0077] FIG. 10 shows normalized mass change .DELTA.m/m.sub.0, where
.DELTA.m and m.sub.0 are the mass change and the initial mass of
the sample, respectively, of the seven as-prepared glasses as a
function of temperature T, the mass change was measured at an
upscanning rate of 10.degree. C./min in air, normalized mass change
.DELTA.m/m.sub.0 of the Cr-containing glass as a function of time t
at T.sub.g=633.degree. C. in air, and
[0078] FIG. 11 shows the onset crystallization temperature T.sub.c
as a function of .DELTA.n.sub.rel,dyn, .DELTA.n.sub.rel,dyn is the
normalized number of moles of the polyvalent element that were
oxidized during dynamic heating in air at 10.degree. C./min to
975.degree. C., T.sub.c was determined from a DSC experiment
performed at 10.degree. C./min in air, and
[0079] FIG. 12 shows SNMS peak area of Ca.sup.2+ for the seven
glasses heat-treated in air at their respective T.sub.g for 6 h as
a function of .DELTA.n.sub.rel,iso, the areas are calculated
between the SNMS concentration curves of Ca.sup.2+ and the
horizontal line through c=c.sub.bulk, .DELTA.n.sub.rel,iso is the
normalized number of moles of the polyvalent element that were
oxidized during iso-thermal heating in air at T.sub.g for 6 h, the
dashed line represents a linear fit, and
[0080] FIG. 13 shows diffusion depths (.DELTA..xi.) of the alkaline
earth ions as a function of the ionic radius r, the glasses have
been heat-treated in H.sub.2/N.sub.2 (1/99) at their respective
T.sub.g for different durations (t.sub.a), closed squares show
t.sub.a=16 h, open squares show t.sub.a=2 h, inset shows plot
of).sup.2 against t.sub.a (0.5, 2, 8, and 16 h) at T.sub.a=T.sub.g
for the Mg-containing glass, and
[0081] FIG. 14 shows Arrhenius plot of In k' as a function of the
reciprocal absolute temperature of the
SiO.sub.2--Na.sub.2O--Fe.sub.2O.sub.3--RO glasses with R.dbd.Mg,
Ca, Sr, Ba heat-treated in H.sub.2/N.sub.2 (1/99) at different
conditions, closed squares: R.dbd.Mg, open squares: R.dbd.Ca,
closed triangles show R.dbd.Sr, open triangles show R.dbd.Ba, inset
shows the corresponding activation energies of diffusion (E.sub.d)
as a function of the ionic radii of the alkaline earth ions (r),
and
[0082] FIG. 15 shows activation energy of diffusion around T.sub.g
(E.sub.d) versus the fragility index m, and the activation energy
of viscous flow at T.sub.g (E.sub..eta.), and
[0083] FIG. 16 shows UV-VIS-NIR spectra of 0.20 mm thick 6 wtFe
glass samples: untreated and heated in H.sub.2/N.sub.2 (1/99) and
CO/CO.sub.2 (98/2) at T.sub.g=926 K for 16 h, and
[0084] FIG. 17 shows SNMS depth profiles of the 6 wtFe glass
heat-treated in CO/CO.sub.2 (98/2) at T.sub.g=926 K for 16 h, the
curves are plotted as concentration of the element at a given depth
divided by the concentration of the same element in the bulk of the
glass (C/C.sub.bulk).
[0085] The present invention will now be described in more detail
in the following.
DETAILED DESCRIPTION OF THE INVENTION
Definitions
[0086] Prior to discussing the present invention in further
details, the following terms and conventions will first be
defined:
Polyvalent Element:
[0087] The term "polyvalent element" can be found in numerous
articles in the field of glass science and technology. In the
present context this term will refer to an element which may exist
in different redox states. The best direct definition one may find
in Pye et al. (2005) p. 28: "In this chapter, all those elements
will be considered to be polyvalent, which may occur in a glass
melt in at least two different oxidation states, even if extreme
oxidizing or reducing conditions are necessary."
Network Modifying Cation:
[0088] The following definition is found in Shelby (2005) p. 10:
"Finally, cations which have very low electronegativities (group
III), and therefore form highly ionic bonds with oxygen, never act
as network formers. Since these ions only serve to modify the
network structure created by the network forming oxides, they are
termed modifiers."
The Glass Transition Temperature (T.sub.g):
[0089] The glass transition temperature (T.sub.g) is defined as the
onset of change of heat capacity due to the glass transition when
heating a glass as defined in Shelby (2005).
[0090] It is a preferred object of the present invention to provide
an improved silicate glass article that solves the above mentioned
problems of the prior art with improved surface properties.
[0091] Cooling of a melt can lead to the formation of either glass
or crystals dependent on the cooling rate [Shelby, 2005]. The
crystalline and glassy materials might have the same composition,
but they differ in structure as crystals have a more ordered
structure than glasses. To discuss the theory of glass formation,
two terms need to be defined: short-range order (SRO) and
long-range order (LRO). SRO exists when the local atomic bonding
units (nearest neighbour configuration of atoms) are uniform in the
entire solid. LRO exists when the arrangement of atoms in space is
periodic [Gersten & Smith, 2001]. Present no ideal definition
of a glass exists. A possible definition is, an amorphous solid
completely lacking LRO, and exhibiting a region of glass
transformation behaviour. The SRO existing in a given glass is
ideally identical to that found in the corresponding crystals.
Crystals are defined to be solids with perfect LRO which implies
perfect periodicity of the atomic arrangement.
[0092] Different theories regarding the structure of oxide glasses
exist, but the random network theory is the most commonly used
model [Shelby, 2005]. The atoms in a glass form a continuous random
network where SRO exists. The conditions for the formation of a
continuous three-dimensional network are [Zachariasen, 1932]:
1) An oxygen atom is linked to not more than two cations. 2) The
oxygen coordination number of the network cation must be small. 3)
The oxygen polyhedra share corners with each others, not edges or
faces. 4) At least three corners in each oxygen polyhedra must be
shared in order to form a three-dimensional network.
[0093] As these rules give no indication of the degree of LRO of
the network, they describe the structure of both glasses and many
crystalline solids. Therefore, an additional requirement was added
by Zachariasen such that the rules only explain glass formation.
The network must be distorted in a way that destroys the LRO. This
distortion can be achieved by variation in bond lengths or angles
as well as by rotations of structural units about their axes
[0094] The chemical components in an oxide glass can be divided
into different categories according to their role in the structural
arrangement of the glass. Stanworth (1971) has classified oxides
into three groups based on the electronegativity of the cation,
i.e., the oxides are classified according to the fractional ionic
character of the cation-anion bond as the anion is oxygen in every
case. If the cation forms bonds with oxygen with a fractional ionic
character near or below 50%, the cation will act as a network
former [Shelby, 2005]. All glasses contain at least one network
former as it is the primary source of the structure.
[0095] In silicate glasses, silicon acts as the network former and
it exists as silicon-oxygen tetrahedral that are linked by bridging
oxygen (BO) atoms. The tetrahedral themselves are very ordered. The
required lack of LRO is introduced by variability in the Si--O--Si
angle, rotation of adjacent tetrahedral around the point occupied
by the oxygen atom linking the tetrahedral, and rotation of the
tetrahedral around the line connecting the linking oxygen with one
of the silicon atoms [Shelby, 2005]. Cations which form highly
ionic bonds with oxygen are termed network modifiers as they only
serve to modify/interfere with the network structure without
becoming part of the primary network [Shelby, 2005]. Network
modifiers provide non-bridging oxygen (NBO) atoms with a negative
charge as they are introduced as oxides and have coordination
number 6. The cations reside in interstitial sites of the network
to maintain local charge neutrality. Both alkali (e.g., Na.sup.+
and K.sup.+) and alkaline earth (e.g., Ca.sup.2+ and Mg.sup.2+)
ions can act as modifiers. Every alkali ion has one neighbouring
NBO, while every alkaline earth ion has two neighbouring NBOs. The
strength of the network is dependent on the amount of network
formers and modifiers. An increase in the amount of network
modifiers results in an increase in the amount of NBOs which
decreases the connectivity (or the degree of polymerization) of
the. This lowers the melting temperature and several other
properties of the glass [Shelby, 2005].
[0096] FIGS. 1A-B show schematic representations of two proposed
mechanisms of surface modification, for explaining the present
invention.
[0097] FIG. 1A shows a known mechanism for the formation of a
crystalline MgO/CaO layer 2 on a silicate glass sample or article 1
comprising Mg.sup.2+, Ca.sup.2+ and Fe.sup.3+. The glass sample 1
is illustrated as having a surface 6, a surface region 3, a bulk
part 4, and a so-called redox front 5. The heat-treatment leads to
oxidation of ferrous iron (Fe.sup.2+) to ferric iron (Fe.sup.3+),
which causes an outward diffusion of divalent cations (primarily
Mg.sup.2+) from the interior of the glass towards the surface. A
crystalline layer 2 forms on the surface 6 as the divalent cations
react with ionic oxygen at the surface. This surface layer 2
exhibits excellent thermal performance.
[0098] FIG. 1B shows a mechanism according to the present invention
for the formation of a silica-rich layer in the surface region 3.
The schematic representation is a still shoot of a dynamic process.
At the redox front 5, the Fe.sup.3+ ions are converted to Fe.sup.2+
ions and electron holes (h.sup. ). The extremely low partial oxygen
pressure in the atmosphere provides a large driving force for the
removal of oxygen from the glass article 1. At the surface, oxygen
anions surrender two electrons to fill the h.sup. and are
subsequently released from the free surface 6 via reaction with
H.sub.2 to form H.sub.2O. The diffusion of h.sup. towards the
surface is charge-balanced by an inward migration of the divalent
cations (including Fe.sup.2+) as a mirror-image of the oxidation
mechanism. Hence, the inward diffusion is driven by reduction of
the high valence to the low valence state of the polyvalent
element. As the network modifying cations (in this example
Mg.sup.2+, Ca.sup.2+ and Fe.sup.3+) leave the surface without the
diffusion of Si.sup.4+ ions, a silica-rich surface layer 3 is
formed. Even though oxygen anions surrender the electrons to h.sup.
at the surface, H.sub.2 molecules in the surrounding atmosphere are
ultimately the source of the electrons.
[0099] Thus, one aspect of the invention relates to a silicate
glass article 1 comprising a bulk part 4 and a surface region 3,
said silicate glass article comprises network-modifying cations
(NMC), e.g. Mg.sup.2+, Ca.sup.2+, and Fe.sup.2+ as indicated in the
FIG. 1B. The concentration of the network-modifying cations (NMC)
in the surface region 3 is lower than in the bulk part 4, and
generally speaking the composition in the surface region of the
network-modifying cations is a consequence of above-mentioned
inward diffusion as it will be explained in more detail below.
[0100] FIG. 2A-D show schematic representations of secondary
neutral mass spectroscopy (SNMS) depth profiles of the original 6
wtFe glass and of the 6 wtFe glass heated in H.sub.2/N.sub.2 (1/99)
at different conditions. The depth profile of each element is
normalised to the bulk concentration. The H.sub.2 partial pressure
is lowered to 0.01 bars in order to create the silica-rich surface
by decreasing the rate of the gaseous permeation. SNMS depth
profiles show that this effort was successful. The chemical
composition of the glasses and various other relevant data is given
in Table 1 below:
TABLE-US-00001 TABLE 1 Chemical composition, iron redox ratio,
density, glass transition temperature (T.sub.g), and NBO/T of the
starting materials. The calculation of NBO/T is explained in the
text below. Glass Chemical composition [wt %] Fe.sup.3+/Fe.sub.tot
Density T.sub.g ID SiO.sub.2 CaO MgO Na.sub.2O Fe.sub.2O.sub.3*
V.sub.2O.sub.5 [at %] [g/cm.sup.3] [.degree. C.] NBO/T 0 wtFe 74.3
10.8 10.2 4.43 -- -- -- 2.501 644 0.84 1 wtFe 72.5 11.4 9.95 4.63
1.09 -- 35-50 2.517 641 0.87 3 wtFe 70.8 11.1 9.59 4.49 3.14 --
60-75 2.549 648 0.84 6 wtFe 69.4 10.8 9.34 4.41 6.07 -- 69 .+-. 3
2.600 653 0.81 1 wtV 72.7 11.4 9.96 4.68 -- 1.03 -- 2.533 653 --
*All iron is reported as Fe.sub.2O.sub.3
[0101] To characterize the network connectivity of the glasses, the
number of non-bridging oxygen atoms per tetrahedron (NBO/T) is
calculated from the chemical compositions. The following formula is
used [Zotov et al., 1992]:
NBO / T = 2 ( [ Na 2 O ] + [ MgO ] + [ CaO ] + [ FeO ] - [ Fe 2 O 3
] ) [ SiO 2 ] + 2 [ Fe 2 O 3 ] ##EQU00001##
where the quantities in the square brackets denote the number of
moles of each oxide. For 1 wtFe and 3 wtFe, the centre values of
the stated iron redox ratio intervals have been used for the
calculation. For the vanadium containing glass, V.sup.5+ is
regarded as a former and V.sup.4+ as a modifier. However, as the
initial vanadium redox ratio has not been determined, the
calculation of NBO/T cannot be performed. For the iron-containing
glasses, T.sub.g increases with increasing glass connectivity
(decreasing NBO/T), which is as expected. The relatively high
T.sub.g of 1 wtV indicates that most of the vanadium is present in
the V.sup.5+ state.
Calculation of NBO/T in Surface Layer
[0102] NBO/T for the untreated 6 wtFe glass is 0.81 as stated in
Table 1. By using the SNMS (concentration of elements), CEMS (redox
state of iron) and FT-IR (concentration of OH groups) data, the
inventors have calculated the NBO/T ratio in a 200 nm surface layer
of the glass treated in H.sub.2/N.sub.2 (1/99 v/v) at T.sub.g for
16 h. This surface layer has NBO/T .about.0.45. This is the only
sample for which the inventors at present have calculated
NBO/T.
[0103] FIG. 2A shows a profile of the original 6 wtFe glass. The
depth profile of the untreated glass reveals that the ion
concentrations do not vary with depth.
[0104] FIG. 2B shows a profile of the 6 wtFe glass heated at
T.sub.g for 2 hours. Heat-treatment of the 6 wtFe glass under a
H.sub.2/N.sub.2 (1/99) gas at T.sub.g, for 2 h leads to the inward
migration of the divalent cations and a remarkable increase in the
silica concentration near the surface.
[0105] 2C shows a profile of the 6 wtFe glass heated at T.sub.g for
16 hours. Thus, by increasing the duration of the heat-treatment it
is possible to increase the thickness of the modified surface layer
as evident from comparison with FIG. 2B.
[0106] 2D shows a profile of the 6 wtFe glass heated at 1.05
T.sub.g for 2 hours. Upon comparison with FIG. 2B, it is seen that
by increasing the temperature of the heat-treatment, the result is
an increase in the thickness of the modified surface layer, i.e.
the depth resulting from the combined heating and reduction
according to the present invention is larger.
[0107] FIG. 3A-D show schematic representations of FT-IR
reflectance spectra of the untreated and heat-treated 6 wtFe glass
at different conditions. When showing the IR reflectance spectra of
the heat-treated samples, only data in the range 900-1200 cm.sup.-1
will be shown as no changes occur at lower wavenumbers. This is
consistent with previous studies as the position and intensity of a
peak at 480 cm.sup.-1 vary little with glass composition.
[0108] FIG. 3A shows spectra of the 6 wtFe glass heated at T.sub.g
for different durations in H.sub.2/N.sub.2 (10/90), and FIG. 3B
shows spectra of the 6 wtFe glass heated for 2 hours at different
temperatures in H.sub.2/N.sub.2 (10/90). For the untreated 6 wtFe
glass, the FT-IR reflectance spectrum displays major peaks near 480
cm.sup.-1 and 1100 cm.sup.-1 that are assigned to Si--O--Si bond
rocking and Si--O--Si antisymmetric stretching vibration,
respectively. With increasing heat-treatment duration (FIG. 3A) and
temperature (FIG. 3B), the following spectral features are
observed. First, the peak at 1100 cm.sup.-1 shifts towards lower
wavenumbers and its intensity decreases. Second, the formation and
growth of a peak at 940 cm.sup.-1 is observed. Third, the formation
and growth of a low-intensity peak near 970 cm.sup.-1 is observed.
The evolution of the Si--O--Si antisymmetric stretching peak shows
that the surface depletes in silica. This confirms corresponding
SNMS results not shown here. The peak at 940 cm.sup.-1 is assigned
to the vibration of Si--OH which is in agreement with the FT-IR
absorption spectroscopy results as OH groups are formed. The weak
peak near 970 cm.sup.-1 is assigned to the vibration of Si--N
bonds.
[0109] FIG. 3C shows spectra of the 6 wtFe glass heated at T.sub.g
for different durations in H.sub.2/N.sub.2 (1/99), and FIG. 3D
shows spectra of the 6 wtFe glass heated for 2 hours at different
temperatures in H.sub.2/N.sub.2 (1/99). The peaks assigned to the
vibration of Si--OH and Si--N bonds are also present in these IR
spectra, but the intensities of the peaks are lower than those
observed for glasses heated in H.sub.2/N.sub.2 (10/90). The
[0110] IR absorption measurements show that less OH groups are
formed with decreasing hydrogen pressure. This explains the lower
intensity of the Si--OH peaks. The Si--O--Si antisymmetric
stretching wavenumber and peak intensity increase with increasing
t.sub.o and T.sub.a. These changes have previously been observed
for silicate glasses when decreasing the total network modifier
content of the surface [Deriano et al., 2004]. This is consistent
with the SNMS results as a silica-rich surface layer is
created.
[0111] FIG. 4A shows a kinetic analysis by a plot of squared
thickness of the of the modified surface region (.DELTA..xi.)
versus heat-treatment duration (t.sub.a) for the 6 wtFe glass
samples in H.sub.2/N.sub.2 (1/99) at T.sub.g. For the reduction
mechanism presented in FIG. 1B to be valid, chemical diffusion of
the divalent cations must rate-limit the reduction kinetics, i.e.,
the diffusion must be parabolic with time. Parabolic kinetics can
be expressed in its integrated form as .DELTA..xi..sup.2=2k't,
where t is time and k' is the parabolic reaction-rate constant.
Hence, the linear relationships found in FIG. 4A prove that the
kinetics signature is indeed parabolic, i.e., it is the diffusion
of the network-modifying cations (NMC; e.g. Mg.sup.2+, Ca.sup.2+,
and Fe.sup.2+) that is rate limiting for the dissipation of the
free energy of the reduction reaction. Clear evidence for the
earlier predicted mechanism (FIG. 1B) has been achieved. Mg.sup.2+
seems to be the fastest diffusing species which is in agreement
with the higher field strength of Mg.sup.2+ compared to Ca.sup.2+
and Fe.sup.2+.
[0112] FIG. 4B shows a table describing the dependence of the
difference in the Fe.sup.2+ concentration, before and after
treatment (.DELTA.c(Fe.sup.2+)), on the initial iron-content of the
glass and the heat-treatment condition. The untreated glasses
contain more Fe.sup.3+ ions with increasing total iron content. As
expected, FIG. 4B reveals that .DELTA.c(Fe.sup.2+) increases with
increasing total iron content of the glass.
[0113] FIG. 4C shows conversion electron Mossbauer spectroscopy
(CEMS) spectra of 6 wtFe glasses (untreated, heated in air at
T.sub.g for 16 h, and heated in H.sub.2/N.sub.2 (1/99) at T.sub.g
for 16 h), the fitted doublets of Fe.sup.3+ and Fe.sup.2+ are
shown. CEMS can be used to study the iron redox state in the
surface region (.about.200 nm) of a sample, i.e., it is different
from conventional Mossbauer spectroscopy that determines the redox
state in the bulk. In conventional Mossbauer spectroscopy, the
absorption peaks of the resonantly absorbed gamma rays are
recorded. In CEMS, the energy released from the excited
(metastable) iron nuclei in the sample is studied. The excited iron
nuclei in the sample return to their ground state by three
processes. Approximately 90% of the absorbed energy is released by
so-called internal conversion and approximately 10% is released as
gamma rays. The internal conversion includes transfer of the energy
via X-rays or to so-called conversion electrons. The conversion
electrons are emitted because the excited nucleus can transfer its
energy to an electron that has a certain probability of being in
the nucleus. In CEMS, the conversion electrons emitted from the
excited nuclei are recorded. These electrons are strongly
attenuated when they pass through the sample, i.e., the signals
only come from the uppermost layer (approximately 200 nm) of the
sample.
[0114] The isomer shifts of Fe.sup.3+ and Fe.sup.2+ are determined
to 0.27.+-.0.06 and 1.07.+-.0.09 mm/s, respectively. The quadrupole
splittings are 1.13.+-.0.09 and 1.7.+-.0.2 mm/s for Fe.sup.3+ and
Fe.sup.2+, respectively. These values are in good agreement with
literature data. The Fe.sup.3+/Fe.sub.tot ratio is estimated for
each sample by measuring the relative areas of the two doublets and
assuming that no metallic iron is present in the glasses. The
calculated ratios are stated in FIG. 4C. The relatively high errors
of the ratios are due to i) the use of a weak source and ii) the
small surface areas of the samples. Fe.sup.3+/Fe.sub.tot equals
68.+-.7% for the untreated 6 wtFe glass. This is consistent with
the result found by conventional Mossbauer spectroscopy that
determined the redox ratio of a powdered sample. As expected,
heat-treatment of the glass in air results in an increased amount
of Fe.sup.3+ ions compared to the amount of Fe.sup.2+ ions near the
surface, whereas the opposite is valid for treatment in
H.sub.2/N.sub.2 (1/99).
[0115] FIG. 5A shows UV-VIS-NIR spectra of 0.20 mm thick 6 wtFe
glass samples heated in H.sub.2/N.sub.2 (1/99) at T.sub.g for
different durations. The iron redox state is investigated as a
function of the different heat-treatment conditions. UV-VIS-NIR
spec-troscopy is the main method used for that purpose. To use this
method quantitatively, the molar absorption coefficient of
Fe.sup.2+ is preliminarily determined.
[0116] FIG. 5B shows the corresponding Fe.sup.2+ concentrations
expressed as the dependence of the difference in the Fe.sup.2+
concentration before and after treatment (.DELTA.c(Fe.sup.2+)) on
the heat-treatment duration (t.sub.a). The increase in the
intensity of the Fe.sup.2+ peak with increasing t.sub.a is less
than the observed increase for glasses heated in H.sub.2/N.sub.2
(10/90) (not shown). Hence, the iron redox ratio is shifted to the
more reduced state with increasing hydrogen partial pressure in the
treatment atmosphere. This is explained by the increased solubility
of H.sub.2 (S.sub.H2) in the glass at higher pressures.
[0117] FIG. 6A shows a table with data of Vickers hardness
(H.sub.v) and water contact angle of the untreated and thermally
treated 6 wtFe glasses at different temperatures and durations. The
Vickers hardness is measured by microindentation. 25 indentations
were performed for each sample at widely separately locations with
a load of 0.25 N and a hold time at the maximum load of 5 s. The
Vickers hardness measurements reveal that the heat-treated glasses
are harder than the original glass. The hardness increases with
duration and temperature of the heat-treatment, i.e., the hardness
increases when the thickness of the modified layer increases. The
contact angle measurements show that the surface becomes more
hydrophobic as a result of the thermal treatments. Hardness
measurements were done with accuracies better than .+-.0.3 GPa.
[0118] FIG. 6B shows a table with data of chemical durability of
the untreated and thermally treated (in H.sub.2/N.sub.2 (1/99)) 6
wtFe glasses at different temperatures and durations. The chemical
resistance of the samples was examined in 0.25 M HCl and 0.25 M KOH
solutions. After immersing a sample in plastic container with the
test solution (20 cm.sup.3 for 1 cm.sup.2 of the glass surface
area), the container was mounted on a thermostatic shaking assembly
at 90.degree. C. (agitated at 100 rpm). After 12 h, the sample was
removed from the solution. The concentrations of leached Na.sup.+
and Mg.sup.2+ ions were measured in the test solution using atomic
absorption spectroscopy (AAnalyst 100, Perkin Elmer). The
dissolution of the glasses was tested in both acid and alkali
solutions. In acidic solutions, primarily the monovalent alkali
ions leave the glass and are replaced by H.sup.+ and/or
H.sub.3O.sup.+. In alkali solutions, the liquid directly attacks
the network bonds as hydroxyl ions can break the Si-0 bonds leading
to the formation of silanolgroups and hence, a continuous
dissolution of the glass. The thermally treated glasses possess a
higher resistance towards both acid and alkali solutions than the
untreated glass (cf. FIG. 6B). The increase in alkali resistance is
caused by the high network connectivity of the treated glasses. The
network modifying cations NMC occupy interstitial positions within
the network creating nonbridging oxygens (NBO). The connected
network on the surface of the treated glasses makes it difficult
for ions to diffuse through the glass, impeding ions such as
OH.sup.- and H.sup.+ to penetrate the network and react with the
glass species. Hence, OH.sup.- diffusion is difficult in the
thermally treated glasses increasing their alkali resistance. The
increase in acid resistance is caused by the impeded diffusion of
H.sup.+ and to a minor extent the depletion of sodium near the
surface in the treated samples.
[0119] In summary, it is possible to create a glass surface
enriched in silica by reducing a polyvalent element present in the
glass. The hardness and chemical durability of the glasses are
increased as a result of the surface modification resulting from
the combined heating and reduction according to the present
invention. This cheap and effective surface modification method can
be used to strengthen any oxide glass containing network modifying
cations (NMC) that can be reduced, e.g. having transition metals,
i.e., glasses possessing properties approaching those of SiO.sub.2
can be created without the requirement to melt the glass at the
high temperatures normally required for silica-rich glasses.
[0120] In the investigated range of T.sub.a and t.sub.a, a change
of T.sub.a or t.sub.a can preferably be used to change the extent
of the surface modification. The heat-treatment atmosphere
determines primarily how the surface is modified in terms of
composition, morphology, and/or redox state. The effects of the
heat-treatment atmosphere on the investigated glass properties are
summarized in Table 2 below:
TABLE-US-00002 TABLE 2 The change in the investigated glass
properties as a function of the heat-treatment atmosphere for the
6wtFe glasses treated at T.sub.g for 16 h. H.sub.2/N.sub.2
H.sub.2/N.sub.2 Property Air N.sub.2 (10/90) (1/99) Stability
against 0 0 -- - crystallization Hardness + 0 -- ++ Crack
resistance 0 - -- ++ Surface hydrophobicity ++ ++ -- ++ Resistance
in acid ++ ++ ++ ++ solution Resistance in alkali -- -- -- ++
solution ++: property increases by more than 5%; +: property
increases by less than 5%; 0: property is unchanged (or within the
error range); -: property decreases by less than 5%; --: property
decreases by more than 5%. All changes are in proportion to the
untreated 6wtFe glass.
[0121] Table 2 may be used to select the appropriate surface
modification method in order to achieve some desired properties.
For most applications of glasses, the effects of treatment in
H.sub.2/N.sub.2 (1/99) are the most favourable.
[0122] FIG. 7 shows a schematic overview of the experimental
strategy and the employed analytical techniques.
EXAMPLES
Impact of Alkali Ions on Formation of SiO.sub.2-Rich Surface
Layer
[0123] The role of alkali ions in the inward cationic diffusion
process is elucidated by answering the following questions: [0124]
What is the impact of the alkali ion on the diffusivity of alkaline
earth ions? [0125] Why are the alkali ions slower than the alkaline
earth ions? [0126] Which alkali ion most effectively creates the
SiO2-rich surface layer?
[0127] To answer these questions, we perform three types of
diffusion experiments.
[0128] First, glasses in the
SiO.sub.2--CaO--Fe.sub.2O.sub.3-A.sub.2O (A=Na, K, Rb, Cs) series
are heat-treated in the reducing H2/N2 (1/99 v/v) atmosphere for a
given duration at various temperatures to determine the activation
energy of Ca2+ diffusion as a function of the type of the alkali
ion.
[0129] Second, heat-treatments are performed at short durations in
order to study the initial phases of the diffusion process.
[0130] Third, glasses with and without alkali and alkaline earth
ions, respectively, are compared in terms of their diffusion
profiles. We also investigate the effect of the alkali ions on the
reduction reactions, density, glass transition temperature, and
fragility of the iron-bearing silicate glasses.
[0131] These results are used to gain insight into the observed
diffusion phenomena.
[0132] We find that under these conditions the Ca.sup.2+ ions
diffuse faster than alkali ions and that the presence of alkali
ions decreases the diffusivity of Ca.sup.2+. In the SiO.sub.2.sup.-
CaO--Fe.sub.2O.sub.3-A.sub.2O (A=Na, K, Rb, or Cs) glass series,
the activation energy of the Ca.sup.2+ diffusion decreases with
alkali size in the sequence Na.sup.+, K.sup.+, Rb.sup.+, and
Cs.sup.+. This trend is coincidence with a decrease of liquid
fragility.
[0133] Sample Preparation. Six glasses (see Table 3) were prepared
from analytical reagent-grade SiO.sub.2, CaCO.sub.3,
Na.sub.2CO.sub.3, K.sub.2CO.sub.3, Rb.sub.2CO.sub.3,
Cs.sub.2CO.sub.3, and Fe.sub.2O.sub.3 powders. The batches were
mixed and melted at 1500.degree. C. in an electric furnace (SF6/17,
Entech) for 3 h in a Pt.sub.90Rh.sub.10 crucible. Afterwards, the
glass melt was quenched on a brass plate and pressed to obtain
cylindrical glasses of 7-10 cm diameter and .about.5 mm height. The
prepared glasses were annealed .about.10 K above their respective
glass transition temperatures for 10 min and then cooled naturally
down to room temperature. It was also attempted to prepare a
lithium-containing glass in the
SiO.sub.2--CaO--Fe.sub.2O.sub.3-A.sub.2O series, but it was not
possible due to phase separation.
TABLE-US-00003 TABLE 3 Chemical Composition, Radius r.sub.A of the
Alkali Ions, Density, Molar Volume, Glass Transition Temperature
T.sub.g, and Fragility Index m of the Prepared Glasses.sup.a Molar
Chemical comp. (mol %) r.sub.A Density volume T.sub.g m Glass A
SiO.sub.2 CaO Fe.sub.2O.sub.3.sup.b A.sub.2O (.ANG.) (g/cm.sup.3)
(cm.sup.3/mol) (K) (--) Si--Ca--Fe--Na Na 67.8 23.3 1.0 7.6 1.02
2.569 23.47 892 41.5 .+-. 0.2 Si--Ca--Fe--K K 68.3 22.9 1.0 7.8
1.38 2.565 24.49 962 37.1 .+-. 0.8 Si--Ca--Fe--Rb Rb 67.9 22.8 1.0
8.0 1.52 2.768 25.41 989 35.3 .+-. 0.4 Si--Ca--Fe--Cs Cs 67.7 23.2
1.0 7.9 1.67 3.011 25.81 1013 34.5 .+-. 0.8 Si--Ca--Fe -- 62.6 36.2
1.1 -- -- 2.672 22.38 1007 n.d..sup.c Si--Na--Fe Na 91.8 -- 1.0 7.0
1.02 2.263 27.08 890 n.d. .sup.ar.sub.A is stated for a
coordination number of 6..sup.8 T.sub.g and m have been determined
by DSC and viscosity measurements, respectively. .sup.bAll iron is
reported as Fe.sub.2O.sub.3. .sup.cn.d.: not determined.
[0134] Cylindrical glass samples (diameter .about.8-10 mm;
thickness 3 mm) were prepared. The samples for the diffusion
experiments were ground flat on one surface to a thickness of
.about.2 mm by a six-step procedure with SiC paper under ethanol.
The surfaces were carefully polished afterwards with 3 .mu.m
diamond paste and finally cleaned with acetone. To study the
reduction reactions, ultraviolet-visible-near-infrared (UV-VIS-NIR)
spectroscopy measurements were performed. The samples for these
experiments were ground coplanar to achieve uniform thickness, and
then they were polished to a thickness of 0.2 mm using the
above-mentioned procedure.
[0135] Sample Characterization. The chemical compositions of the
glasses are listed in Table 3. They were analyzed by x-ray
fluorescence (XRF) on a S4-Pioneer spectrometer (Bruker-AXS). The
main impurity in the glasses was Al.sub.2O.sub.3 (.about.0.2 mol
%). Densities of the glasses were measured by He-pycnometry
(Porotech) and are also shown in Table 3.
[0136] The glass transition temperature (T.sub.g) was measured
using a differential scanning calorimetry (DSC) instrument (STA
449C Jupiter, Netzsch). The isobaric heat capacity (C.sub.p) curve
for each measurement was calculated relative to the C.sub.p curve
of a sapphire reference material after subtraction of a correction
run with empty crucibles. Measurements were carried out in a purged
Ar atmosphere. The following heating procedure was carried out to
determine T.sub.g. First, the sample was heated at 10 K/min to a
temperature .about.1.11 times the respective T.sub.g (in K) of each
sample. Subsequently, the sample was cooled to room temperature at
10 K/min. Then, T.sub.g was determined by a second upscan at 10
K/min in order to ensure a uniform thermal history of the glasses.
T.sub.g was defined as the cross point between the extrapolated
straight line of the glass C.sub.p curve before the transition zone
and the tangent at the inflection point of the sharp rise curve of
C.sub.p in the transition zone.
[0137] Viscosity was measured by beam-bending (T>T.sub.g) and
concentric cylinder (T>T.sub.liquidus) experiments. For
beam-bending experiments, bars of 45 mm length and 3.times.5
mm.sup.2 cross-section were cut from the bulk glasses. The bars
were bent in a symmetric 3-point forced bending mode with 40 mm
open span (VIS 401, Bahr). A 300 g weight was used to explore the
viscosity range from approximately 1012 to 10.sup.9.5 Pas at a
constant heating rate of 10 K/min. The viscosity was calculated
according to DIN ISO 7884-4. The low viscosities (<10.sup.3 Pas)
were measured using a concentric cylinder viscometer. The
viscometer consisted of furnace, viscometer head, spindle, and
sample crucible. The viscometer head (Physica Rheolab MC1, Paar
Physica) was mounted on top of a high temperature furnace (HT 7,
Scandiaovnen A/S). Spindle and crucible were made of
Pt.sub.80Rh.sub.20. The viscometer was calibrated using the
National Bureau of Standards (NBS) 710A standard glass.
[0138] Thermal Treatment. To induce the reduction reactions and
diffusion processes, the polished glasses were heat-treated at 1
atm in an electric furnace under a flow of H.sub.2/N.sub.2 (1/99
v/v) gas. The glass samples were inserted into the cold furnace and
the gas-flow was turned on. The furnace was then heated at 10 K/min
to the pre-determined heat-treatment temperature and kept at this
temperature for a given duration. Afterwards, the furnace was
cooled down to room temperature at 10 K/min.
[0139] The glasses in the SiO.sub.2--CaO--Fe.sub.2O.sub.3-A.sub.2O
(A=Na, K, Rb, Cs) series were treated at 0.95, 1.00, 1.025, and
1.05 times their respective T.sub.g (in K) for 2 h, at their
respective T.sub.g for 60 h, and at the T.sub.g (892 K) of the
Si--Ca--Fe--Na glass for 60 h. Additionally, the Si--Ca--Fe--Na
glass was treated at its T.sub.g for 0.2, 1, 8, and 16 h,
respectively. The ternary Si--Ca--Fe and Si--Na--Fe glasses were
treated at their respective T.sub.g for 2 h.
[0140] UV-VIS-NIR Spectroscopy. Usually, iron in glasses exists in
the states of Fe.sup.2+ and Fe.sup.3+. In this work, UV-VIS-NIR
absorption spectroscopy was used to determine the change in the
valence state of iron of 0.20 mm thick samples for the
SiO.sub.2--CaO--Fe.sub.2O.sub.3-A.sub.2O (A=Na, K, Rb, Cs) series.
UV-VIS-NIR spectra were recorded over the wavelength range of 300
to 1100 nm using a UV-VIS-NIR Specord 200 spectrophotometer
(Analytik Jena AG) at a resolution of 1 nm. The spectra were
recorded with air as reference.
[0141] The ferrous (Fe.sup.2+) ion has a broad absorption peak with
maximum absorbance at 1050-1100 nm. The position and maximum
absorbance of this peak varies with glass composition and the
absorption coefficients for our glasses were not known. Therefore,
the absorption coefficient of the Lambert-Beer equation for the
Si--Ca--Fe--Na glass was calculated: A=c.epsilon.x, where A is
absorbance, c the concentration, .English Pound. the absorption
coefficient, and x the sample thickness. In another study, the
redox state [Fe.sup.3+]/[Fe.sub.tot], where
[Fe.sub.tot]=[Fe.sup.2+]+[Fe.sup.3+], of this glass was found by
Mossbauer spectroscopy to be 77.+-.2 at %. By using the
[Fe.sup.3+]/[Fe.sub.tot] ratio and the total iron content, the
concentration of ferrous iron was calculated in the untreated
Si--Ca--Fe--Na glass. Plotting the absorbance near 1075 nm versus
the sample thickness (0.12, 0.20, 0.40, and 0.80 mm) gave a linear
relation (R.sup.2=0.995). From the slope of this plot (c.epsilon.),
the absorption coefficient was calculated to be 3.82 L mol.sup.-1
mm.sup.-1.
[0142] Secondary Neutral Mass Spectroscopy. To investigate the
diffusion processes, compositional analysis of the surfaces was
carried out using electron-gas secondary neutral mass spectroscopy
(SNMS). The measurements were performed on an INA3 (Leybold AG)
instrument equipped with a Balzers QMH511 quadrupole mass
spectrometer and a Photonics SEM XP1600/14 amplifier. The analyzed
area had a diameter of 5 mm and was sputtered using Kr plasma with
an energy of .about.500 eV. The time dependence of the sputter
profiles was converted into depth dependence by measuring the depth
of the sputtered crater at 12 different directions on the same
sample with a Tencor P1 profilometer.
[0143] Results. FIG. 8 show a dependence of viscosity .eta. on
temperature T for the SiO.sub.2--CaO--Fe.sub.2O.sub.3-A.sub.2O
glasses with A=Na, K, Rb, Cs. An increase in viscosity with
increasing ionic radius r.sub.A of the alkali ion at a given
temperature is observed for both the low and high temperature data.
To determine the liquid fragility index, we fit the viscosity data
with the Mauro-Yue-Ellison-Gupta-Allan (MYEGA) equation,
log .eta. = log .eta. .infin. + ( 12 - log .eta. .infin. ) T g T
exp [ ( m 12 - log .eta. .infin. - 1 ) ( T g T - 1 ) ] ,
##EQU00002##
where .eta..sub..infin., T.sub.g, and m are fitting parameters.
.eta..sub..infin. is the viscosity at infinite temperature and m is
the fragility index of the glass-forming liquid. The fragility
index is defined as the slope of the log .eta. versus T.sub.g/T
curve at T.sub.g:
m .ident. log .eta. ( T g / T ) T = T g . ##EQU00003##
[0144] In the model, the viscosity at T.sub.g is set equal to
10.sup.12 Pas since this has been shown to be equivalent to the
calorimetrically measured T.sub.g values for oxide glasses. The
MYEGA equation offers improved accuracy in performing low
temperature extrapolations compared to the Vogel-Fulcher-Tammann
(VFT) and Avramov-Milchev (AM) equations. We apply the above
equation in fitting the viscosity data. The fitting is done using a
Levenberg-Marquardt algorithm. The fitted values of m are shown in
the inset of FIG. 8 as a function of r.sub.A. For the glass melts
in the series SiO.sub.2--CaO--Fe.sub.2O.sub.3-A.sub.2O with A=Na,
K, Rb, Cs, the fragility decreases with increasing size of the
alkali ion. In contrast, the glass transition temperature increases
with r.sub.A.
[0145] UV-VIS-NIR spectra of untreated and heat-treated glasses in
the SiO.sub.2--CaO--Fe.sub.2O.sub.3-A.sub.2O (A=Na, K, Rb, Cs)
series are not shown. However, a maximum absorption peak is found
at approximately 1075 nm, which is due to the presence of Fe.sup.2+
ions. The maximum absorbance and position of the Fe.sup.2+ peak is
the same (within .+-.5%) for all glasses. This indicates that the
initial [Fe.sup.3+]/[Fe.sub.tot] ratio is approximately the same in
all the glasses. This has been additionally confirmed by performing
thermogravimetric measurements on the glasses. The mass increase
due to incorporation of oxygen was the same (within .+-.6%) for all
six glasses. By using Mossbauer spectroscopy, we have found that
the untreated Si--Ca--Fe--Na glass contains 77.+-.2% of its Fe ions
as Fe.sup.3+. To study the kinetics of the reduction reaction, the
Si--Ca--Fe--Na glass has been heat-treated at its T.sub.g of 892 K
for 2, 8, 16, and 60 h. With increasing treatment duration t.sub.a,
the absorbance of the Fe.sup.2+ band increases because Fe.sup.3+ is
reduced to Fe.sup.2+. The change in Fe.sup.2+ concentration
increases approximately linearly with the square root of the
treatment duration, implying that diffusion-controlled kinetics
occurs. UV-VIS-NIR spectra of the
SiO.sub.2--CaO--Fe.sub.2O.sub.3-A.sub.2O (A=Na, K, Rb, Cs) glasses
heat-treated at their respective T.sub.g for 60 hours and at the
T.sub.g of the Si--Ca--Fe--Na glass (892 K) for 60 hours are not
shown. When the glasses are treated at the same temperature the
change in absorbance of the Fe.sup.2+ peak increases with
increasing radius of the alkali ion. When the glasses are treated
at their respective T.sub.g, the trend is qualitatively the same
but the differences between the glasses are more pronounced. This
is because the Si--Ca--Fe--Cs glass has the highest T.sub.g and
therefore it is treated at the highest temperature.
[0146] The SNMS technique has been employed to study the
reduction-induced diffusion processes in the six glasses. The
method provides information about the surface composition of the
glass as a function of depth. Depth profiles are not shown.
However, it is found that the depth profile of the Si--Ca--Fe--K
glass that has been heat-treated in H.sub.2/N.sub.2 (1/99) for 2 h
at 1.05T.sub.g=1010 K. A depletion of calcium, potassium, and iron
is observed near the surface. The extent of the calcium depletion
is larger than that of potassium and iron. Qualitatively all six
glasses display the same type of surface depletion of
network-modifying cations as a result of heat-treatment for 2 h at
temperatures around their respective T.sub.g. The surface depletion
is caused by an inward diffusion of these ions induced by the
reduction of Fe.sup.3+ to Fe.sup.2+. An important consequence of
the inward diffusion is the creation of a silica-rich surface
layer. Before heat-treatment in H.sub.2/N.sub.2 (1/99), the glasses
do not show any variation in composition as a function of depth. To
study the initial phases of the diffusion process, the
Si--Ca--Fe--Na glass has been heat-treated at its T.sub.g for 1
hour and 0.2 hours. The glass treated for 0.2 h displays an about
50 nm layer depleted in calcium and iron, whereas the
concentrations of silicon and sodium are higher in this layer than
in the bulk. When the duration of the treatment is increased to 1
h, the thickness of the layer depleted in calcium and iron
increases and inward diffusion of sodium occurs. Furthermore, an
enrichment of sodium is observed in the depth interval from
approximately 50 to 100 nm. The ternary Si--Ca--Fe and Si--Na--Fe
glasses have been heat-treated at their respective T.sub.g for 2 h.
Inward diffusion of Ca.sup.2+ and Na.sup.+, respectively, is also
observed in these glasses.
[0147] For the SiO.sub.2--CaO--Fe.sub.2O.sub.3-A.sub.2O (A=Na, K,
Rb, Cs) glass series, the temperature dependence of the calcium
diffusion has been systematically investigated. To quantitatively
analyze the effect of the alkali ion on the diffusion of Ca.sup.2+,
the diffusion depth (.DELTA..xi.) of Ca.sup.2+ is calculated as the
first depth at which c/c.sub.bulk 1 for three measurements in
succession. The inward cationic diffusion of alkaline earth ions is
parabolic with time, thus we calculate the rate constant k' of the
calcium diffusion:
k ' = ( .DELTA. .xi. ) 2 t d , ##EQU00004##
where t.sub.d is the diffusion time (2 h). k' is proportional to
the product of the diffusion coefficient of the rate-limiting
species (divalent cation) and a thermodynamic driving force
(gradient in oxygen activity). Hence, from the temperature
sensitivity of k', the activation energy of calcium diffusion
(E.sub.d) can be obtained by plotting the data in Arrhenius
coordinates (FIG. 9). The diffusion data for each glass reveal an
Arrhenius dependence on temperature in the studied temperature
range. E.sub.d is calculated from the slope of each line and is
plotted as a function of the ionic radius of the alkali ion.
E.sub.d decreases with increasing size of the alkali ion.
[0148] Effect of Alkali Ion on Redox-Diffusion Processes. When the
glasses in the SiO.sub.2--CaO--Fe.sub.2O.sub.3-A.sub.2O (A=Na, K,
Rb, Cs) series are heat-treated at the T.sub.g of the
Si--Ca--Fe--Na glass (i.e., at the same temperature), the order of
the degree of reduction follows the same trend as that of the molar
volume of these glasses (Table 3). Since the modifier ions are
believed to occupy interstitial sites in the network, they block
the paths for the small H.sub.2 molecules. Hence, when the glass
structure is relatively open, it is easier for the H.sub.2
molecules to permeate the glass. In comparison, the diffusion data
show that the isothermal inward diffusion of Ca.sup.2+ is fastest
in the Si--Ca--Fe--Na glass, i.e., in the glass with the lowest
molar volume. This is because two simultaneous processes contribute
to the reduction of Fe.sup.3+ to Fe.sup.2+: H.sub.2 permeation and
outward flux of electron holes. The former process dominates the
reduction reaction at 0.01 bar of H.sub.2. Therefore, the thickness
of modified surface layer (as measured by SNMS) cannot be directly
correlated with the degree of reduction (as measured by UV-VIS-NIR
spectroscopy). In other words, a large extent of Fe.sup.3+
reduction does not necessarily result in a thick SiO.sub.2-rich
surface layer because two processes contribute to the reduction of
Fe.sup.3+.
[0149] There must therefore be another reason for why large alkali
ions cause the slowest isothermal Ca.sup.2+ diffusion in the
glasses. The Si--Ca--Fe--Cs glass has the highest T.sub.g of the
glasses and it has been shown that this type of redox-induced
diffusion begins at temperatures around 0.8T.sub.g (in K).
Apparently, the process requires some degree of viscous softening
even though the motion of the alkaline earth ions is decoupled from
that of the network. Therefore, the glass with lowest T.sub.g will
have the fastest Ca.sup.2+ diffusion when the glasses are
heat-treated at the same temperature.
[0150] The temperature sensitivity of the Ca.sup.2+ diffusion shows
that the diffusion activation energy (E.sub.d) decreases with
increasing r.sub.A. In another study, we predicted the presence of
interconnected channels in the glass network based on the modified
random network (MRN) model of Greaves. We found that the alkaline
earth diffusion can be enhanced by lowering the liquid fragility
due to the more simple diffusion paths in strong systems. This
could also explain the results in this study since there is a
positive correlation between E.sub.d and m in the
SiO.sub.2--CaO--Fe.sub.2O.sub.3-A.sub.2O (A=Na, K, Rb, Cs)
series.
[0151] Regarding the inward diffusion, the results obtained in this
study clearly demonstrate that alkali ions diffuse slower than the
divalent calcium ions. In the beginning of the process, the alkali
ions have not started to diffuse to any significant extent and the
diffusion of Ca.sup.2+ and Fe.sup.2+ completely dominates. The high
concentration of sodium in the surface layer of these glasses is
caused by the inward diffusion of calcium and iron. The Ca.sup.2+
diffusion is faster than that of Na.sup.+ since the extent of
Ca.sup.2+ diffusion in the Si--Ca--Fe--Na glass is larger than that
of Na.sup.+ in the ternary Si--Na--Fe glass. The two glasses are
comparable since they have approximately the same T.sub.g.
Furthermore, the presence of alkali ions decreases the diffusivity
of the alkaline earth ions because the extent of Ca.sup.2+
diffusion is smaller in the Si--Ca--Fe--Cs glass than in the
ternary Si--Ca--Fe glass. These two glasses also have approximately
the same T.sub.g. The presence of relatively slow alkali ions may
therefore block the diffusion of the faster alkaline earth ions in
the interconnected channels, i.e., the slow alkali ions occupy
interstices and hereby increase the packing density. These
interstices can no longer be used for alkaline earth migration.
[0152] The reason for why the alkaline earth ions are faster than
the alkali ions must be that the divalent alkaline earth ions are
the most suitable ones for carrying the positive charge that
charge-balance the outward flux of electron holes. An alkaline
earth ion neutralizes two electron holes, whereas an alkali ion
neutralizes only one electron hole. In addition, the alkaline earth
ions are more mobile than trivalent modifier ions (e.g., Al.sup.3+)
because the latter ones are more strongly bound to the oxygen
anions. The diffusion coefficient of the alkali ions is smaller in
the inward diffusion process compared to what is found in the
literature because the latter results have predominantly been
obtained by the use of a radioactive tracer.
[0153] Conclusions. We have studied the impact of alkali ions on
the diffusion of calcium ions in the glass transition range in
iron-bearing silicate glasses. The diffusion is induced by
thermally treating the glasses in a reducing atmosphere at
temperatures around T.sub.g. This treatment causes a reduction of
Fe.sup.3+ to Fe.sup.2+, which requires an inward diffusion of
mobile cations. We have found that the mobility of Ca.sup.2+
strongly depends on the type of the alkali ion present in the glass
and that the diffusion of Ca.sup.2+ is faster than that of the
alkali ions. The presence of alkali ions decreases the mobility of
Ca.sup.2+ and the activation energy of Ca.sup.2+ diffusion
decreases with increasing radius of the alkali ion. The latter
trend is coincidence with a decrease of liquid fragility and an
increase of glass transition temperature.
Impact of Polyvalent Element on Formation of SiO2-Rich Surface
Layer
[0154] In this work, seven different polyvalent elements (Fe, Mn,
Cu, Ce, Ti, V, and Cr) are examined in terms of their impact on the
formation of the silica-rich surface layers. This is done in order
to understand the mechanism of the internal diffusion and to
broaden and optimizing the application of the methods. As a result
of this work, it can be found out which element is the most
suitable ingredient of the glass for obtaining the crystalline
oxide layer and the silica-rich layer, respectively.
[0155] Seven soda-lime silicate glasses, each of which contains one
of the following polyvalent metals: Fe, Mn, Cu, Ce, Ti, V, and Cr,
are oxidized in air and reduced in H.sub.2/N.sub.2 (1/99) at their
respective glass transition temperatures for some period. A
crystalline oxide surface layer is created on the glasses (except
the vanadium-bearing glass) under the oxidizing condition, since
the metallic ions are oxidized from lower to higher valence state,
and thereby calcium ions diffuse outward and react with oxygen
ions. In contrast, a silica-rich surface layer is created on the
glasses under the reducing condition, since sodium and calcium ions
diffuse inward. It is found that the extent of both outward and
inward diffusions strongly depends on the type of the polyvalent
ions for the same conditions of heat-treatment. Out of the seven
polyvalent metals studied in this work, copper induces the highest
extent of both the inward and outward diffusion, and hence, the
thickest surface layer of both amorphous silica and crystalline
alkaline earth oxides. The oxide layer lowers the onset temperature
of the primary crystallization. The silica-rich surface layer
enhances the chemical resistance of the glass in a hot basic
solution.
[0156] Sample preparation. Seven glasses were prepared from three
main analytical reagent-grade chemicals (SiO.sub.2,
Na.sub.2CO.sub.3, and CaCO.sub.3) and one minor analytical
reagent-grade polyvalent metal oxide (Cr.sub.2O.sub.3, MnO.sub.2,
CeO.sub.2, V.sub.2O.sub.5, CuO, Fe.sub.2O.sub.3, or TiO.sub.2). The
batch materials were melted in a Pt.sub.90Rh.sub.10 crucible in an
electric furnace (SF6/17, Entech) at 1500.degree. C. for 3 h. The
melt was then cast onto a brass plate and pressed to obtain
cylindrical glasses of 7-10 cm diameter and .about.5 mm height. The
prepared glasses were immediately annealed at 640.degree. C. for 10
min and then cooled naturally down to room temperature in the
closed furnace. The chemical compositions of the glasses were
analyzed by x-ray fluorescence (S4-Pioneer, Bruker-AXS) and are
listed in Table 4. The main impurity was Al.sub.2O.sub.3 (<0.1
mol %).
TABLE-US-00004 TABLE 4 Chemical composition and glass transition
temperature (T.sub.g) of the prepared glasses containing various
polyvalent oxides (A.sub.xO.sub.y). Polyvalent Chemical composition
(mol %) oxide SiO.sub.2 Na.sub.2O CaO A.sub.xO.sub.y T.sub.g
(.degree. C.) TiO.sub.2 65.6 8.7 23.7 1.9 636 V.sub.2O.sub.5 68.1
7.8 23.1 0.9 630 Fe.sub.2O.sub.3 67.8 7.6 23.3 1.0 619 CuO 66.5 7.8
23.5 2.0 621 Cr.sub.2O.sub.3 67.2 8.0 22.6 2.1 633 CeO.sub.2 67.3
7.9 22.8 2.0 635 MnO.sub.2 66.6 8.1 23.0 2.1 621
[0157] Thermal analyses. The glass transition temperature (T.sub.g)
was measured using differential scanning calorimetry (DSC). The DSC
measurements were performed on a simultaneous thermal analyser
(STA) (449C Jupiter, Netzsch). All the glasses were subjected to
two runs of up- and downscans at 10.degree. C./min. The onset
temperature of the endothermic C.sub.p (isobaric heat capacity)
jump during the second upscan was assigned as T.sub.g.
[0158] The STA instrument was also used for recording both DSC and
thermogravimetric (TG) signals during both iso-thermal (i.e,
constant temperature) and dynamic (i.e., increasing temperature at
constant heating rate) heating, from which the oxidation degree of
the polyvalent ions were determined. The measurements were
performed on powdered samples by crushing and sieving the glass
samples. The 45-63 .mu.m size fraction was collected for each
glass. A platinum crucible containing the glass sample and an empty
platinum crucible were placed on the sample carrier of the STA at
room temperature. To evaporate water from the samples, the
crucibles were initially heated at a rate of 10.degree. C./min to
300.degree. C. and held for 15 min before cooling down to room
temperature. For iso-thermal heating experiments, both crucibles
were then held 5 min at an initial temperature of 60.degree. C. and
heated at a rate of 10.degree. C./min to the respective T.sub.g of
the glasses and held at this temperature for 6 or 12 h. For dynamic
heating experiments, the crucibles were also held 5 min at an
initial temperature of 60.degree. C. but were then heated at a rate
of 10.degree. C./min to 975.degree. C. After both types of heating
schedules had ended, the crucibles were cooled down to 250.degree.
C. at a rate of 10.degree. C./min, and finally down to room
temperature at a natural rate. Atmospheric air dried by a molecular
sieve was used as purge gas.
[0159] Determination of redox states of polyvalent elements. By
using a ultraviolet-visible-near-infrared (UV-VIS-NIR) Specord 200
spectrophotometer (Analytik Jena AG) at a resolution of 1 nm,
UV-VIS-NIR spectra were recorded over the wavelength range of
300-1100 nm. The measurements were performed on 2.0 mm thick
samples ground by a six-step procedure with SiC paper, followed by
polishing with 3 .mu.m diamond suspension. The UV-VIS-NIR spectra
were used to qualitatively determine the redox states of the
polyvalent elements present in the untreated glasses.
[0160] Post-treatment and characterizations. The bulk glasses were
cut in cylinders of 10 mm diameter and 2-3 mm height. One surface
of each sample was then ground by a six-step procedure with SiC
paper, followed by polishing with 3 .mu.m diamond suspension. To
induce the inward cationic diffusion, the polished glasses were
heat-treated at 1 atm in an electric furnace under a flow of
H.sub.2/N.sub.2 (1/99 v/v) gas. The glass samples were inserted
into the cold furnace and the gas-flow was turned on. The furnace
was then heated at 10.degree. C./min to the respective T.sub.g of
the glasses and kept at this temperature for 6 h. The diffusion
process was ended by cooling the furnace to room temperature at
10.degree. C./min. To induce the outward cationic diffusion, the
polished glasses were heat-treated by applying an identical heating
procedure under atmospheric conditions.
[0161] The diffusion profiles were determined by electron-gas
secondary neutral mass spectroscopy (SNMS). SNMS is used to
determine the elemental concentrations as a function of the depth
within the glass. The measurements were performed by using an INA3
(Leybold AG) instrument equipped with a Balzers QMH511 quadrupole
mass spectrometer and a Photonics SEM XP1600/14 amplifier. The
analyzed area had a diameter of 5 mm and was sputtered using Kr
plasma with an energy of .about.500 eV. The time dependence of the
sputter profiles was converted into depth dependence by measuring
the depth of the crater at 12 different locations on the same
sample with a Tencor P1 profilometer.
[0162] Chemical durability test. The chemical durability of both
untreated and heat-treated samples was measured in a hot basic
solution to see the impact of the internal diffusion on glass
properties. The chemical durability of the samples was examined in
0.25 M KOH solution (pH=13.2). After immersing a sample in a
plastic container with the test solution (20 cm.sup.3 for 1
cm.sup.2 of the glass surface area), the container was mounted on a
thermostatic shaking assembly at 90.degree. C. (agitated at 100
rpm). After 6 h, the sample was removed from the solution. The
concentration of leached Ca.sup.2+ ions was measured in the test
solution using atomic absorption spectroscopy (AAnalyst 100, Perkin
Elmer).
[0163] Results and discussion. UV-VIS-NIR spectra was made (not
shown) of the untreated Cr- and Ti-containing glasses. Cr can exist
as Cr.sup.2+, Cr.sup.3+, and Cr.sup.6+ in silicate glasses. No
absorption bands due to Cr.sup.2+ are observed, whereas both
Cr.sup.3+ (at 445, 640, 660, and 690 nm) and Cr.sup.6+ (at 360 nm)
are observed. Ti can exist as Ti.sup.4+ and Ti.sup.3+. Ti.sup.4+
has d.sup.0 electron configuration, in which only charge transfer
transitions occur in the UV range. Hence, Ti.sup.4+ is colorless
and no absorption bands can be observed in the spectrum. It has
been reported that Ti.sup.3+ can have a single band centered at 570
nm, but this is not observed in this work. Fe and Mn can exist as
Fe.sup.2+ and Fe.sup.3+ and Mn.sup.2+ and Mn.sup.3+, respectively.
The di- and trivalent states are detected in both glasses. V can
exist as V.sup.3+, V.sup.4+, and V.sup.5+, but only V.sup.4+ is
detected. V.sup.5+ is expected to be present in silicate glasses
melted in air, but its strong charge transfer bands are found in
the UV-range, which causes a sharp UV-absorption edge. Cu can exist
as Cu.sup.0, Cu.sup.+, and Cu.sup.2+, and Cu.sup.2+ is observed in
the spectrum of the untreated glass, whereas Cu.sup.+ is colorless.
Finally, Ce can exist as Ce.sup.3+ and Ce.sup.4+. Both redox states
cause absorption peaks in the UV-range, and hence, they cannot be
observed due to a sharp UV-absorption edge. Table 4 exhibits the
chemical compositions and the T.sub.g values of the seven glasses
studied in this work. As expected, the glasses with the higher
field strength polyvalent metal ions (Ti, Ce, Cr, V) have higher
values of T.sub.g than the rest. For example, Ti.sup.4+ is known to
act as a network former and V.sup.5+ may serve to increase the
polymerization degree of a silicate network.
[0164] To study the impact of the polyvalent element on the
oxidation process and the crystallization behavior of the
as-produced glasses, both DSC and TG measurements were carried out
in air at a heating rate of 10.degree. C./min on powdered samples.
The energy response of the samples to the dynamic heating and the
mass change of the glasses during the heating are measured using
DSC and TG, respectively. An increase in mass of 0.20% at
temperatures above 550.degree. C. was observed. The increase in
mass is caused by oxidation of Cr.sup.2+ to Cr.sup.3+ and/or
Cr.sup.3+ to Cr.sup.6+. However, because Cr.sup.2+ is only present
in very small amounts in silicate glasses prepared under oxidizing
melting conditions, and it is not observed in the UV-VIS-NIR
spectrum, the oxidation of Cr.sup.2+ to Cr.sup.3+ can be neglected.
According to the oxidation mechanism of iron, the oxidation
reaction causes the incorporation of oxygen into the glass by
forming metallic surface oxides. The oxidation of Cr.sup.3+ to
Cr.sup.6+ is connected with a weak exothermic peak in the DSC curve
that has its maximum at .about.705.degree. C. between the glass
transition temperature (T.sub.g=633.degree. C.) and the onset
temperature of crystallization (T.sub.c=855.degree. C.). The
inflection point of the TG curve (not shown) corresponds to the
maximum of the oxidation peak.
[0165] The TG traces of the seven glasses heated at 10.degree.
C./min in air are displayed in FIG. 10. All glasses display a mass
increase similar to that of the Cr-containing glass, i.e., all the
studied polyvalent elements have ions that can be oxidized.
However, the extent of the mass increase differs between the
glasses. To calculate the number of moles .DELTA.n of each
polyvalent element that has been oxidized, the stoichiometry of the
oxidation reaction must be known. According to Table 4, the
following reactions take place when the Cr- and Mn-containing
glasses are heated above T.sub.g:
4Cr.sup.3++3O.sub.2-->4 Cr.sup.6++6O.sup.2-
4 Mn.sup.2++O.sub.2-->4 Mn.sup.3++2O.sup.2-
[0166] The stoichiometry of the second reaction is also valid for
oxidation of Ti.sup.3+ to Ti.sup.4+, V.sup.4+ to V.sup.5+,
Fe.sup.2+ to Fe.sup.3+, Cu.sup.+ to Cu.sup.2+, and Ce.sup.3+ to
Ce.sup.4+. It is assumed that the mass increase of the samples
during heating is solely due to incorporation of oxygen. Mossbauer
spectroscopy experiments have confirmed this assumption for the
oxidation of Fe.sup.2+ to Fe.sup.3+ during heating of iron-bearing
aluminosilicate glass fibers in air. The normalized number of moles
that has been oxidized (.DELTA.n.sub.rel) can then be calculated by
the following equation:
.DELTA. n rel = x .DELTA. m m 0 M O 2 ##EQU00005##
where m.sub.0 is the initial mass of the sample, .DELTA.m is the
maximum increase in mass, and M.sub.O.sub.2 is the molar mass of
oxygen. x is the ratio between the number of moles of the
polyvalent element being oxidized and the number of moles of
O.sub.2 being consumed in the oxidation process, i.e., x is 4/3 for
oxidation of Cr.sup.3+ to Cr.sup.6+ and x is 4 for the other
oxidation reactions. The values calculated using the equation for
the dynamic heating procedure (.DELTA.n.sub.rel,dyn) are listed in
Table 5. Cu.sup.+ is oxidized to the largest extent, whereas very
limited amounts of Ti.sup.3+ and V.sup.4+ are oxidized.
TABLE-US-00005 TABLE 5 .DELTA.n.sub.rel, .sub.iso,
.DELTA.n.sub.rel, dyn, T.sub.c, and T.sub.p of the as-produced
glasses. Polyvalent .DELTA.n.sub.rel, iso .DELTA.n.sub.rel, dyn
element (.mu.mol/g) (.mu.mol/g) T.sub.c (.degree. C.) T.sub.p
(.degree. C.) Ti 8 10 888 915 V 5 15 893 922 Fe 66 93 871 892 Cu 80
105 820 841 Cr 63 83 855 890 Ce 41 62 884 907 Mn 59 84 869 888
.DELTA.n.sub.rel, iso and .DELTA.n.sub.rel, dyn are the normalized
number of moles of the polyvalent element that were oxidized during
iso-thermal heating in air at T.sub.g for 6 h and during dynamic
heating in air at 10.degree. C./min to 975.degree. C.,
respectively. The characteristic temperatures were determined from
DSC measurements performed at an upscanning rate of 10.degree.
C./min in air.
[0167] Table 5 shows the T.sub.c (onset temperature of
crystallization) and T.sub.p (peak temperature of crystallization)
values of the glasses. In general, the crystallization begins at a
lower temperature when a larger mass increase occurs, i.e., a
higher degree of oxidation during the dynamic heating (FIG. 11).
During the DSC upscanning in air, oxidation of the lower valence
state ions takes place, which leads to formation of an oxide
surface layer. By using x-ray diffraction (XRD), atomic force
microscopy (AFM), and secondary neutral mass spectroscopy (SNMS),
it has been proved that the oxide layer is nano-crystalline. The
results obtained in this study indicate that the nano-crystalline
layer lowers the activation energy for crystallization since the
crystals can grow from the nuclei that are already formed at the
surface. This agrees with the results of previous studies, in which
it has been shown by DSC experiments that the crystallization
starts at the surface when the nano-crystalline surface layer is
present.
[0168] The mass increase during iso-thermal heating of the powdered
samples in air at their respective T.sub.g for 6 h was also
measured by TG. In the inset of FIG. 10, the mass change of the
Cr-containing glass is shown. The normalized number of moles
oxidized during the iso-thermal heating procedure
(.DELTA.n.sub.rel,iso) is calculated using before mentioned
equation and listed in Table 5. There is a positive correlative
between .DELTA.n.sub.rel,iso and .DELTA.n.sub.rel,dyn, i.e., when a
large degree of oxidation occurs as a result of the dynamic
heating, it also occurs as a result of the iso-thermal heating.
[0169] To study the diffusion processes associated with the
oxidation and reduction reactions, SNMS is used to determine the
concentration depth profiles of the seven glasses that prior to the
measurements have been oxidized in air or reduced in
H.sub.2/N.sub.2 (1/99) at their respective T.sub.g for 6 h. It
should be noticed that the untreated glasses show no changes in
composition as a function of depth. The normalized concentration
depth profiles of Si, O, Ca, and Na of the Cu-containing glass
oxidized in air are not shown. A high surface concentration of
calcium is found, which is due to outward diffusion of Ca.sup.2+.
The low surface concentration of sodium and silicon is due to the
enrichment of calcium and oxygen near the surface. The
concentration profiles of the Cu-containing glass as a result of
heat-treatment in H.sub.2/N.sub.2 (1/99) are not shown. A depletion
of sodium and calcium is found near the surface, which causes
formation of the high silica concentration in the surface layer.
The uncertainty in the detection of the polyvalent elements by SNMS
is relatively high because of their low concentration. Therefore,
it has not been possible to evaluate whether these elements have
diffused as a result of the heat-treatments.
[0170] Outward and inward diffusion of Ca.sup.2+ occurs in all the
glasses when they are heated in air and H.sub.2/N.sub.2 (1/99),
respectively. The only exception is the V-containing glass, in
which no outward diffusion is observed as a result of the
heat-treatment in air. This must be due to the very limited
oxidation of V.sup.4+ (Table 5). To quantitatively compare the
degree of Ca.sup.2+ diffusion for the different glasses, the peak
area (A.sub.Ca) of the Ca.sup.2+ curve and the diffusion depth
(D.sub.Ca) of Ca.sup.2+ are calculated. The areas are calculated
between the SNMS concentration curve of Ca.sup.2+ and the
horizontal line through c=c.sub.bulkD.sub.Ca,ox and D.sub.Ca,red
are calculated as the first depth at which c/c.sub.bulk.ltoreq.1
and c/c.sub.bulk.ltoreq.1, respectively, for three measurements in
succession. The calculated values are listed in Table 6. A positive
correlative between A and D is found.
TABLE-US-00006 TABLE 6 SNMS peak areas and diffusion depths of
Ca.sup.2+ for the seven glasses heat-treated under oxidizing (air)
and reducing (H.sub.2/N.sub.2 1/99) atmospheres at their respective
T.sub.g for 6 h. Polyvalent A.sub.Ca, ox D.sub.Ca, ox A.sub.Ca, red
D.sub.Ca, red element (nm) (nm) (nm) (nm) Ti 0.7 81 3.1 93 V 0 0
7.8 249 Fe 2.3 220 10.5 264 Cu 5.4 307 13.1 315 Cr 2.4 235 7.1 202
Ce 1.7 153 4.6 173 Mn 2.1 209 2.6 83 The areas are calculated
between the SNMS concentration curves of Ca.sup.2+ and the
horizontal line through c = c.sub.bulk.
[0171] To study if the degree of oxidation of a given polyvalent
element is linked with the degree of diffusion, A.sub.Ca,ox is
plotted as a function of .DELTA.n.sub.rel,iso (FIG. 12). Both
A.sub.Ca,ox and .DELTA.n.sub.rel,iso were obtained from the
iso-thermal heating procedure, but it should be noticed that
A.sub.Ca,ox was determined by using a bulk sample, whereas
.DELTA.n.sub.rel,iso was determined by using a powdered sample. If
the mass changes were obtained by using bulk samples, the mass
increase would have been below the detection limit of the apparatus
due to the small surface area. FIG. 12 shows that the degree of
Ca.sup.2+ diffusion increases approximately linearly with
increasing degree of oxidation of the polyvalent element. This
clearly demonstrates that the outward diffusion is driven by the
oxidation of the polyvalent element. A similar tendency is expected
between the degree of inward diffusion and the degree of reduction
of the polyvalent element. However, this has not been possible to
investigate because the TG measurements could not be conducted in
the reducing H.sub.2/N.sub.2 atmosphere due to use of platinum
crucibles and sample holder.
[0172] Based on the results represented in Table 6, it is concluded
that Cu is the element for creating the thickest layers under the
same heat-treatment conditions. The elements that are either highly
reduced or oxidized before the heat-treatments create the thinnest
modified surface layers. This could be explained as follows. For
example, if an element is fully reduced before heat-treatment in
H.sub.2/N.sub.2 (1/99), the concentration of ions that may be
reduced is low, and hence, the created layer will be thin. On the
other hand, if the element is almost fully oxidized, the
concentration of reducible ions is high, but the reduction reaction
is thermodynamically unfavorable, which will also result in a thin
layer.
[0173] The impact of the surface modifications on the glass
properties is investigated by determining the chemical durability
of both untreated and heat-treated glasses in 0.25 M KOH solution
(pH=13.2). The concentrations of Ca.sup.2+ in the leaching
solutions after 6 h at 90.degree. C. are given in Table 7.
TABLE-US-00007 TABLE 7 Chemical durability of the as-produced
glasses and the glasses heat- treated in air and H.sub.2/N.sub.2
(1/99) at their respective T.sub.g for 6 h. Polyvalent
c(Ca.sup.2+).sub.unt c(Ca.sup.2+).sub.ox c(Ca.sup.2+).sub.red
element (mg/L) (mg/L) (mg/L) Ti 2.5 2.6 2.3 V 2.6 2.5 1.8 Fe 2.6
2.9 1.6 Cu 2.5 3.1 1.4 Cr 2.6 2.6 1.8 Ce 2.6 2.7 2.0 Mn 2.7 2.8 2.2
Chemical durability is expressed by the leached amount of Ca.sup.2+
(c(Ca.sup.2+)) ions after 6 h in 0.25M KOH solution at 90.degree.
C. Concentration measurements were done with accuracies better than
.+-.0.2 mg/L.
[0174] During the leaching experiment, the hydroxyl ions directly
attack the Si--O network bonds, resulting in the formation of
silanol groups (--Si--OH):
##STR00001##
[0175] A continuous dissolution of the glass from the surface is
the result of this process. The untreated glasses do not have
significantly different chemical durabilities. Some of the glasses
that have been oxidized in air display a lower resistance towards
the basic solution than the corresponding untreated glasses. This
is explained by the low surface concentration of silicon because
few Si--O bonds need to be broken in order to dissolve a relatively
large amount of Ca.sup.2+ ions. The increase in basic resistance of
the glasses reduced in H.sub.2/N.sub.2 (1/99) is caused by the high
network connectivity of the treated glasses due to the inward
diffusion of Ca.sup.2+. It seems that for both samples heated in
air and H.sub.2/N.sub.2 (1/99), the chemical durability depends on
the thickness of the modified surface layer.
[0176] Conclusions. Oxidation and reduction of seven polyvalent
elements in silicate glasses result in diffusion processes near the
surface. The oxidation process leads to formation of a crystalline
oxide surface layer, whereas the reduction process leads to
formation of a silica-rich layer. The crystalline surface layer
lowers the onset temperature of the primary crystallization
process, whereas the silica-rich surface layer enhances the
chemical resistance of the glass in a hot basic solution. The
diffusion mechanisms of modifying ions appear to be universal for
all polyvalent element-containing glasses at temperatures around
T.sub.g. To create the thickest modified surface layer, the
polyvalent element must be present in the glass as a mixture of
oxidized and reduced ions. Of the studied elements, Cu is the
optimal ingredient for formation of the thickest surface layers
under the same redox treatment condition.
Impact of Alkaline Earth Ions on Formation of SiO2-Rich Surface
Layer
[0177] Here, we investigate the influence of the type of the
alkaline earth ion on the inward diffusion process in the
SiO.sub.2--Na.sub.2O--Fe.sub.2O.sub.3--RO (R.dbd.Mg, Ca, Sr, Ba)
glass series. We also attempt to find out whether and how the
inward cationic diffusion is correlated with the viscous flow
behaviour of glasses in the glass transition range, i.e., with the
liquid fragility. The latter is a generally accepted concept that
describes the extent of the non-Arrhenius flow. The liquid
fragility is related to the glass composition and structure and the
glass structure strongly influences the energy barrier of the
diffusion of electron holes and modifying ions. In the present
work, both a kinetic fragility index m (i.e., steepness of the log
viscosity vs. T.sub.g/T curve at T.sub.g) and a thermodynamic index
C.sub.pl/C.sub.pg (i.e., the ratio of the liquid to the glassy
isobaric heat capacity at T.sub.g) are determined as measures of
liquid fragility. Finally, we also study the influence of the
alkaline earth ion on the redox state of iron using Mossbauer
spectroscopy because the inward cationic diffusion process is
affected by the initial Fe.sup.3+ concentration.
[0178] We have studied the correlation between liquid fragility and
the inward diffusion (from surface towards interior) of alkaline
earth ions in the SiO.sub.2--Na.sub.2O--Fe.sub.2O.sub.3--RO
(R.dbd.Mg, Ca, Sr, Ba) glass series. The inward diffusion is caused
by reduction of Fe.sup.3+ to Fe.sup.2+ under a flow of
H.sub.2/N.sub.2 (1/99 v/v) gas at temperatures around the glass
transition temperature (T.sub.g). The consequence of such diffusion
is the formation of a silica-rich nanolayer. During the reduction
process, the extent of diffusion (depth) decreases in the sequence
Mg.sup.2+, Ca.sup.2+, Sr.sup.2+ and Ba.sup.2+, whereas the
fragility increases in the same sequence. It is found that the
ratio of the activation energy of the inward diffusion E.sub.d near
T.sub.g to the activation energy for viscous flow E.sub..eta. at
T.sub.g increases with increasing fragility of the liquid. The
inward cationic diffusion can be enhanced by lowering the fragility
of glass systems via varying the chemical composition.
[0179] Sample preparation. Four iron-bearing alkali-alkaline-earth
silicate glasses (see Table 8) were prepared from analytical
reagent-grade SiO.sub.2, Na.sub.2CO.sub.3, MgO, CaCO.sub.3,
SrCO.sub.3, BaCO.sub.3, and Fe.sub.2O.sub.3 powders. The mixed
batch materials were melted in an electrical furnace (SF6/17,
Entech) at 1500.degree. C. in a Pt.sub.90Rh.sub.10 crucible for 3
h. The melt was then cast onto a brass plate and pressed to obtain
cylindrical glasses of 7-10 cm diameter and .about.5 mm height. The
prepared glasses were annealed 10 K above their respective glass
transition temperatures for 10 min and then cooled down to room
temperature within 20 h.
TABLE-US-00008 TABLE 8 Chemical composition, density, molar volume
(=molar mass/density), and iron redox ratio of the prepared
glasses. The radii r of the alkaline earth ions are stated for a
coordination number of 6. Chemical Molar Fe.sup.3+/ composition
(mol %) Density volume Fe.sub.tot r R SiO.sub.2 Na.sub.2O
Fe.sub.2O.sub.3* RO (g/cm.sup.3) (cm.sup.3/mol) (at %) (.ANG.) Mg
69.0 7.7 1.1 21.7 2.475 20.772 74 .+-. 2 0.72 Ca 67.8 7.6 1.0 23.3
2.569 23.469 77 .+-. 2 1.00 Sr 68.6 7.7 1.0 22.4 3.022 23.498 80
.+-. 5 1.18 Ba 66.8 8.1 1.0 23.2 3.295 25.212 74 .+-. 2 1.35 *All
iron reported as Fe.sub.2O.sub.3
[0180] Sample characterization. The chemical compositions of the
glasses were analyzed by X-ray fluorescence (S4-Pioneer,
Bruker-AXS) and are listed in Table 8. The main impurity was
Al.sub.2O.sub.3 (.about.0.2 mol %). Densities of the glasses were
measured by He-pycnometry (Porotech) and are also given in Table 8.
Transmission .sup.57Fe Mossbauer spectroscopy on powdered samples
was employed to study the effect of the alkaline earth ion on the
redox state of the Fe ions. A constant acceleration spectrometer
with a source of .sup.57Co in rhodium was used and calibrated with
.alpha.-Fe. Measurements were made at room temperature and data
were collected for one week for each sample. Isomer shifts are
given relative to that of the calibration spectrum.
[0181] The glass transition temperature (T.sub.g) was measured
using a differential scanning calorimetry (DSC) instrument (STA
449C Jupiter, Netzsch). The C.sub.p curve for each measurement was
calculated relative to the C.sub.p curve of a sapphire reference
material after subtraction of a correction run with empty
crucibles. Measurements were carried out in a purged Ar atmosphere.
The following heating procedure was carried out to determine
T.sub.g. First, the sample was heated at 10 K/min to a temperature
1.11 times the respective T.sub.g (in K) of each sample.
Subsequently, the sample was cooled to room temperature at 10
K/min. Then, T.sub.g was determined by a second upscan at 10 K/min
in order to ensure a uniform thermal history of the four glasses.
The ratio C.sub.pl/C.sub.pg was also determined from this scan. To
determine the liquid fragility, the viscosity was measured by
beam-bending (T>T.sub.g) and concentric cylinder
(T>T.sub.liquidus) experiments. For beam-bending experiments,
bars of 45 mm length and 3.times.5 mm.sup.2 cross-section were cut
from the bulk glasses. The bars were blended in a symmetric 3 point
forced bending mode with 40 mm open span (VIS 401, Bahr). A 300 g
weight was used to explore the viscosity range from approximately
10.sup.12 to 10.sup.10 Pas at a constant heating rate of 10 K/min.
The viscosity was calculated according to DIN ISO 7884-4. The low
viscosities (<10.sup.2 Pas) were measured using a concentric
cylinder viscometer. The viscometer consisted of four parts:
furnace, viscometer head, spindle, and sample crucible. The
viscometer head (Physica Rheolab MC1, Paar Physica) was mounted on
top of a high temperature furnace (HT 7, Scandiaovnen A/S). Spindle
and crucible were made of Pt.sub.80Rh.sub.20. The viscometer was
calibrated using the National Bureau of Standards (NBS) 710A
standard glass.
[0182] Diffusion experiments. As diffusion depths below 1 .mu.m are
expected based on the above studies, the sample surfaces were
carefully prepared. First, the bulk glasses were cut in cylinders
of 10 mm diameter and 2-3 mm height. One surface of each sample was
then ground by a six-step procedure with SIC paper, followed by
polishing with 1 .mu.m diamond suspension.
[0183] To induce the inward cationic diffusion, the polished
glasses were heat-treated at 1 atm in an electrical furnace under a
flow of H.sub.2/N.sub.2 (1/99 v/v) gas. The presence of oxygen in
the furnace is not completely avoidable. But it is possible to keep
its partial pressure at a known value by using a
Fe.sub.3O.sub.4/Fe.sub.2O.sub.3 redox buffer. Fe.sub.2O.sub.3 and
Fe.sub.3O.sub.4 powders were mixed in the molar ratio 3:2 and
placed inside the furnace together with the samples. The glass
samples and redox buffer were inserted into the cold furnace and
the gas-flow was turned on. The furnace was then heated at 10 K/min
to the pre-determined heat-treatment temperature T.sub.a and kept
at this temperature for the duration t.sub.a. The diffusion process
was ended by cooling the furnace to room temperature at 10 K/min.
The glasses were treated at 0.95, 1.00, 1.025, and 1.05 times their
respective T.sub.g (in K) for 2 h and at their T.sub.g for 16 h. In
addition, the Mg-containing glass was treated at its T.sub.g for
0.5 and 8 h. The diffusion profiles were determined by electron-gas
secondary neutral mass spectroscopy (SNMS). SNMS is used to
determine the elemental concentrations as a function of the depth
within the glass. The measurements were performed by using an INA3
(Leybold AG) instrument equipped with a Balzers QMH511 quadrupole
mass spectrometer and a Photonics SEM XP1600/14 amplifier. The
analyzed area had a diameter of 5 mm and was sputtered using Kr
plasma with an energy of .about.500 eV. The time dependence of the
sputter profiles was converted into depth dependence by measuring
the depth of the crater at 12 different locations on the same
sample with a Tencor P1 profilometer.
[0184] Results. The transmission .sup.57Fe Mossbauer spectrum of
the untreated Ca-containing glass at 295 K is not shown. The two
doublets, seen in the spectrum, are due to paramagnetic Fe.sup.3+
(isomer shift of 0.28 mm/s and quadrupole splitting of 1.07 mm/s)
and Fe.sup.2+ (isomer shift of 1.00 mm/s and quadrupole splitting
of 1.86 mm/s). A sextet due to Fe.sup.3+ is also seen in the
spectrum. It can be due to the presence of clusters that may have
formed during quenching. The sextet and the two doublets appear in
the Mossbauer spectra of all the four glasses, only the areas of
the peaks vary. The relative spectral areas of Fe.sup.3+ (doublet
and sextet) and Fe.sup.2+ (doublet) are used to calculate the
Fe.sup.3+/Fe.sub.tot ratio for each of the untreated glasses (Table
8).
[0185] The isobaric heat capacity (C.sub.p) curves recorded during
DSC upscans for the four glass compositions are not shown. T.sub.g
is determined at the cross point between the extrapolated straight
line of the glass C.sub.p curve before the transition zone and the
tangent at the inflection point of the sharp rise curve of C.sub.p
in the transition zone. The T.sub.g values within an accuracy of
.+-.3 K are given in Table 9. T.sub.g is plotted as a function of
the ionic radius of the alkaline earth ion (r) (not shown). The
radii of the alkaline earth ions are listed in Table 8 for a
coordination number of 6. T.sub.g is found to decrease with
increasing r.
TABLE-US-00009 TABLE 9 Characteristic parameters of the
SiO.sub.2--Na.sub.2O--Fe.sub.2O.sub.3--RO glasses with R = Mg, Ca,
Sr, Ba determined by DSC and viscosity measurements. T.sub.g
C.sub.pl - C.sub.pg R (K) C.sub.pl/C.sub.pg (J g.sup.-1 K.sup.-1) F
m Mg 912 1.22 0.28 2.68 .+-. 0.02 35.7 .+-. 0.4 Ca 892 1.27 0.32
3.00 .+-. 0.05 39.4 .+-. 0.8 Sr 858 1.29 0.35 3.30 .+-. 0.09 42
.+-. 2 Ba 823 1.30 0.36 3.55 .+-. 0.07 45 .+-. 1 The errors of F
and m are within the 95% confidence limits.
[0186] The viscosity data for the four compositions obtained from
beam-bending and concentric cylinder viscometry are not shown.
However, for both the low and high temperature data, a decrease in
viscosity (n) with increasing r is observed. To describe the
temperature dependence of viscosity of glass-forming liquids, we
apply the Avramov-Milchev (AM) equation:
log .eta. = A + B ( T g T ) F ##EQU00006##
where A, B, F, and T.sub.g are constants, which are obtained by
fitting the viscosity data to the equation. A is log
.eta..sub..infin., where .eta..sub..infin. is the viscosity at
infinite temperature. F is the fragility index of the glass-forming
liquid and it is a function of the chemical composition at ambient
pressure. The higher the F value, the more fragile the liquid.
Since the viscosity for oxide glasses at T.sub.g is equal to
10.sup.12 Pas, the equation may be simplified as the following
expression:
log .eta. = log .eta. .infin. + ( 12 - log .eta. .infin. ) ( T g T
) F . ##EQU00007##
[0187] In this equation, there are only 3 parameters. The data are
fitted with this modified AM equation by using the
Levenberg-Marquardt algorithm. It is found that this equation fits
the data better than both the Vogel-Fulcher-Tammann (VFT) equation
and the Adam-Gibbs (AG) equation. The F values for the four
compositions are given in Table 8. Fragility can also be described
by the index m, which is defined as the slope of the log .eta.
versus T.sub.g/T curve at T.sub.g:
m = log .eta. ( T g / T ) T g . ##EQU00008##
[0188] F can be converted to m through the following relation:
m=(12-log .eta..sub..infin.)F.
[0189] The calculated values of m are listed in Table 9. The
studied glass melts become more fragile with increasing size of the
alkaline earth ion. The fragility described here is the so-called
kinetic fragility. Several attempts have been made to correlate the
kinetic fragility with thermodynamic property changes at T.sub.g.
It has been suggested to use the ratio of the heat capacity of the
liquid to that of the glass at T.sub.g (C.sub.pl/C.sub.pg) as a
measure of thermodynamic fragility. We have calculated
C.sub.pl/C.sub.pg for the four compositions based on the DSC
measurements and the values are stated in Table 9. C.sub.p, is the
offset value of the C.sub.p overshoot above the glass transition
range. To determine C.sub.pg, a linear function is fitted to the
C.sub.p values at temperatures below T.sub.g. The value of this
function at T.sub.g is reported as C.sub.pg. The results in Table 9
show that C.sub.pl/C.sub.pg increases with increasing r. For
comparison, the step change in the heat capacity
(C.sub.pl/C.sub.pg) at the glass transition is also calculated and
listed in Table 9. Similar tendencies for C.sub.pl/C.sub.pg and
C.sub.pl/C.sub.pg are observed.
[0190] The diffusion profiles in the four glasses heat-treated
under H.sub.2/N.sub.2 (1/99) at their respective T.sub.g for 16 h
are not shown. However, all glasses display a depletion of sodium,
iron, and the respective alkaline earth ion near the surface. This
inward diffusion causes the high surface concentration of silica.
It should be noted that the untreated glasses show no changes in
composition as a function of depth. To quantitatively analyze the
data, the diffusion depths (.DELTA..xi.) of the alkaline earth ions
are determined for T.sub.a=T.sub.g (FIG. 12). .DELTA..xi. is
calculated as the first depth at which c/c.sub.bulk.ltoreq.1 for
three measurements in succession. FIG. 12 shows that the order of
the alkaline earth ion mobility at isokom temperatures is:
Mg.sup.2+>Ca.sup.2+>Sr.sup.2+>Ba.sup.2+.
[0191] For the reduction mechanism to be valid, chemical diffusion
of the alkaline earth ions must occur, i.e., the diffusion must be
parabolic with time. Parabolic kinetics can be expressed in its
integrated form as:
(.DELTA..xi.).sup.2=k't,
where t is time and k' is a constant. k' is proportional to the
product of the diffusion coefficient of the rate-limiting species
and the normalized (to RT) thermodynamic driving force. A kinetic
analysis is performed by plotting (.DELTA..xi.).sup.2 against the
duration of the heat-treatment (0.5, 2, 8, and 16 h) at
T.sub.a=T.sub.g for the Mg-containing glass. A linear relationship
is found with a coefficient of determination (R.sup.2) of 0.999
(see inset of FIG. 13). This proves that the kinetic signature is
indeed parabolic.
[0192] To study the temperature dependence of the alkaline earth
diffusion, we calculate the temperature sensitivity of k' from the
change of .DELTA..xi. with temperature at a constant diffusion
time. FIG. 14 presents the resulting Arrhenius plots. The diffusion
data for each glass reveal an Arrhenius dependence on temperature
(see solid lines in FIG. 14). From the slope of each line, an
activation energy of diffusion around T.sub.g (E.sub.d) is
calculated and shown as a function of r in the inset of FIG. 14.
E.sub.d increases with increasing r, and hence, with decreasing
field strength of the alkaline earth ions.
[0193] Since only the nature of the alkaline earth ion has been
changed and not its concentration, the number of non-bridging
oxygens (NBOs) is the same in all glasses when neglecting the small
variations in the compositions of the glasses (Table 8). As it is
known, variation of the fragility of silicate melts may result from
small changes of hydroxyl content. However, differences in hydroxyl
content are predominantly due to variations of the melting
conditions, which are not modified for the silicate glasses of this
study. Thus, we assume only marginal changes in hydroxyl
concentration of the untreated glasses. In addition, the iron redox
ratio does not differ between the samples when considering the
error range of the data (see Table 8), which is in agreement with
the findings in a previous study. Hence, the observed changes in
T.sub.g, fragility, and diffusion cannot be explained by the
concentration of NBOs or Fe.sup.3+. The changes must be due to the
difference in size of the alkaline earth ions (and hence, in their
field strength), in ionic packing density, and bond angle
distributions.
[0194] The increase in T.sub.g with decreasing r of alkaline earth
ions (R.sup.2+) is attributed to the strengthening of the overall
network since a decrease of r also leads to an increase in their
field strength, and hence, to an enhanced attraction of R.sup.2+
ions to their surrounding structural groups of [SiO.sub.4]
tetrahedra. The Mg.sup.2+ ions most strongly attract the nearby
[SiO.sub.4] tetrahedra, and hence, a higher potential energy
barrier needs to be overcome to initiate glass transition.
Similarly, the viscosity decreases at both high and low
temperatures with increasing r. In accordance with the alkaline
earth field strength also the molar volume of the glasses decreases
in the sequence Ba>Sr>Ca>Mg, indicating a less open
structure for high field strength cations (Table 8).
[0195] The kinetic fragility (quantified by F or m) shows a
positive correlation with the thermodynamic fragility (quantified
by C.sub.pl/C.sub.pg or C.sub.pl/C.sub.pg). However, it has been
shown that this correlation is not generally true for small organic
and polymeric liquids, whereas the correlation exists for inorganic
glass-forming liquids. The fragility is found to increase with
increasing r of the alkaline earth ions in the glass series studied
in this work. This may be explained as follows. For a more fragile
liquid, there is a larger change in the structure of the liquid
with temperature than for a less fragile liquid. The high field
strength of Mg.sup.2+ causes a high degree of short range order,
which prevents the structure from a rapid break-down with
increasing temperature.
[0196] Our diffusion experiments have shown that the Mg.sup.2+ ions
are the fastest at isokom temperatures and have the lowest E.sub.d.
It has previously been reported that alkaline earth ions are most
mobile in alkali alkaline earth silicate glasses when the radii of
the alkali and alkaline earth ions are similar. The jump of an
alkaline earth ion from one octahedral site to another leaves
behind a high negative charge density that induces an electric
dipole moment. This moment might cause a backward jump of the
alkaline earth ion. However, when the alkali and alkaline earth
ions have similar radii, the highly mobile alkali ions can easily
enter the alkaline earth ion sites and hereby reduce the
probability of the backward jump. In our sodium alkaline earth
silicate glasses, the Ca.sup.2+ ions should then be the fastest and
have the lowest E.sub.d as the radius of Na.sup.+ (1.02 .ANG.) is
very similar to that of Ca.sup.2+ (1.00 .ANG.). In addition, the
iron reduction causes the alkali ions to diffuse (role of Na.sup.+
in diffusion process is discussed later). These factors limit the
ability of Na.sup.+ to jump into the empty alkaline earth ion
sites, which might explain why Ca.sup.2+ is not found to be the
fastest ion in our glasses.
[0197] According to the MRN model, the network modifying oxides
form interconnected channels (i.e., a percolative network) at
sufficiently high concentration. The threshold for percolation
occurs at 16 vol % of modifying oxides, which is exceeded by the
glass compositions studied in this work. The alkaline earth ions
should diffuse fastest through the channels when their size is
smallest which explains our diffusion results at isokom
temperatures. The activation energy of diffusion is the sum of an
electrostatic term due to the Coulomb interaction between the
cation and the NBO plus an elastic part to open up doorways into
neighboring sites. In our glasses, the latter term governs the
activation energy as the smallest alkaline earth ion has the lowest
E.sub.d since it most easily moves through the channels. The
channels are constituted by [SiO.sub.4] tetrahedra, i.e., the
required displacement of oxygen is relatively small for a small
alkaline earth ion.
[0198] To study the link between ionic diffusion and fragility,
E.sub.d is plotted against m in FIG. 15. m is found to be
proportional to E.sub.d, implying that the diffusion of alkaline
earth ions in glasses is related to the liquid fragility. This
could be explained as follows. Strong glass systems have a smaller
configurational entropy (S.sub.c) than fragile glass systems.
S.sub.c is the part of the entropy of a pure liquid that is
determined by the abundance of possible packing states obtainable
at the temperature T. Therefore, fragility depends on the
multiplicity of states (local potential energy minima), i.e.,
strong systems will have a structure with fewer accessible states
than fragile systems. Alkali and alkaline earth ions should
therefore diffuse faster in strong systems than in fragile systems
due to the simple diffusion paths in the former ones.
[0199] To study if the diffusion of alkaline earth ions in glasses
is linked to the viscous flow of the network, the following
relation is considered:
E.sub..eta.=mT.sub.gR ln10=(12-log .eta..sub..infin.)FT.sub.gR
ln10,
where E.sub..eta. is the activation energy of viscous flow at
T.sub.g and R is the gas constant. E.sub.d is plotted against
E.sub..eta. in FIG. 15 and a clear linear correlation is observed,
but the E.sub.d/E.sub..eta. ratio is smaller than 1. According to
the Stoke-Einstein equation, the activation energy of diffusion
increases with increasing viscosity. However, the equation cannot
be used to predict ion mobilities, because the ions use the
transportation route with the lowest activation energy, i.e., they
flow faster than the cooperative rearrangements of the structural
units. In other words, the diffusion of alkali and alkaline earth
ions is decoupled from the network change.
[0200] The role of Na.sup.+ in the diffusion processes appears to
be complex. Even though alkali ions are normally considered to be
the fastest ions in glasses, the diffusion depth of Na.sup.+ is
generally smaller than that of the alkaline earth ions, which is in
agreement with our previous studies. In addition, an enrichment
peak of Na.sup.+ (compared to the surrounding Na.sup.+
concentration) is found near the surface of the heat-treated
samples for R.dbd.Ca, Sr, and Ba, but not for R.dbd.Mg (not shown).
The height and width of this peak decrease with decreasing T.sub.a
and t.sub.a. These results imply that Na.sup.+ diffuses back to the
surface after its initial inward diffusion. This agrees with the
existence of an interdiffusion mechanism of alkaline earth ions and
sodium ions as suggested by the values of the decoupling ratios.
These issues will be addresses in more detail in a future study by
investigating the reduction induced diffusion in 68 SiO.sub.2-23
CaO-8 R.sub.2O (R.dbd.Na, K, Rb, Cs)-1 Fe.sub.2O.sub.3 glasses.
[0201] Other features of the inward cationic diffusion process are
discussed in the following. The diffusion profiles (not shown)
indicate that the diffusion of Ca.sup.2+, Sr.sup.2+, and Ba.sup.2+
occurs in different step, whereas this is not the case for
Mg.sup.2+. For the Mg-containing glass, the change in concentration
with depth is approximately linear. The inward diffusion of
Ca.sup.2+, Sr.sup.2+, and Ba.sup.2+ might be slowed down at larger
depths due to the accumulation of these relatively large ions. This
would explain why the depth, at which the slope of the
concentration versus depth curve suddenly changes, seems to
decrease with increasing r.
[0202] The inward diffusion process is driven by reduction of
Fe.sup.3+ to Fe.sup.2+, but Fe.sup.2+ is capable of diffusing
itself. The ionic radius of Fe.sup.2+ in 6-fold coordination is
0.78 .ANG. for the high spin state. The diffusion data of iron
reveal two general features. First, the diffusion depth of
Fe.sup.2+ decreases with increasing r. Second, the ratio between
the diffusion depth of Fe.sup.2+ and that of the alkaline earth ion
(.DELTA..xi..sub.Fe/.DELTA..xi..sub.R) decreases with increasing r
(in average: 0.9 for R.dbd.Mg and 0.4 for R.dbd.Ba). These
observations are explained by the steric hindrance induced by the
larger alkaline earth ions on the diffusion of Fe.sup.2+.
[0203] Conclusions. By means of an inward diffusion process driven
by reduction of iron, we have studied the diffusion of alkaline
earth ions in the glass transition range in silicate glasses and
the link to the liquid fragility. The fragility is found to
increase with increasing ionic radius of the alkaline earth ion.
The diffusion is studied by heat-treating the glasses at
temperatures around T.sub.g as this causes an inward diffusion of
mobile cations due to reduction of Fe.sup.3+ to Fe.sup.2+. The
determined activation energies of diffusion (E.sub.d) reveal that
the small alkaline earth ions are the most mobile and that E.sub.d
increases with increasing fragility. We have explained our results
based on the modified random network model, which predicts the
formation of percolation channels in the studied glasses. The small
ions most easily move through these channels that are constituted
by [SiO.sub.4] tetrahedra. The results imply that the inward
cationic diffusion can be enhanced by lowering the fragility of
glass systems. In accordance with the diffusion mechanism, it is
found that E.sub.d<E.sub..eta. as the alkaline earth ions bypass
the slow cooperative rearrangements of the glass network by using
the transportation path with the lowest activation energy. The
inward cationic diffusion process can be used to create a
silica-rich nanolayer on glass surfaces and the results obtained in
this study show that Mg.sup.2+ ions most effectively creates this
layer at isokom temperatures.
Impact of Reducing Gas on Formation of SiO2-Rich Surface Layer
[0204] To examine whether a gas with larger reducing molecules than
H.sub.2 at a higher pressure in the atmosphere can be used to
induce the formation of the silica-rich layer, we apply a
CO/CO.sub.2 (98/2 v/v) atmosphere as a reducing agent for
heat-treatments at T.sub.g of iron-bearing silicate glasses.
Afterwards, we compare the concentration profile of the surface
layer of the CO/CO.sub.2 treated glass with that of the
H.sub.2/N.sub.2 (1/99) treated glass. The information on the
dependence of the creation of the silica-rich surface layer on the
gas type and composition is important for clarifying the mechanism
of the inward diffusion process and for the application of the
surface modification technology.
[0205] We find that inward diffusion of network-modifying cations
can occur in an iron-containing silicate glass when it is
heat-treated in CO/CO.sub.2 (98/2 v/v) or H.sub.2/N.sub.2 (1/99
v/v) gases at temperatures around the glass transition temperature.
The inward diffusion is caused by the reduction of ferric to
ferrous ions and this diffusion leads to formation of a silica-rich
surface layer with a thickness of 200.about.600 nm. The diffusion
coefficients of the network-modifying divalent cations are
calculated and they are different for the glasses treated in the CO
and H.sub.2 gases. At the applied partial pressures of CO and
H.sub.2, the H.sub.2-bearing gas creates the silica-rich layer more
effectively than the CO-bearing gas. The layer increases the
hardness and chemical durability of the glass due to the silica
network structure in the surface layer.
[0206] Experimental. Two glasses named 6 wtFe and 3 wtFe were
prepared by melting mixtures of analytical reagent-grade raw
materials at 1500.degree. C. under atmospheric air. Composition of
the 6 wtFe glass (in wt %) is 69.4 SiO.sub.2, 10.8 CaO, 9.3 MgO,
4.4 Na.sub.2O, and 6.1 Fe.sub.2O.sub.3, whereas that of 3 wtFe
glass is 71.0 SiO.sub.2, 11.1 CaO, 9.6 MgO, 4.5 Na.sub.2O, and 3.2
Fe.sub.2O.sub.3. Here, all iron (Fe.sup.2+ and Fe.sup.3+) is
reported as Fe.sub.2O.sub.3. NaO and CaO were introduced into the
batch using their respective carbonates. SiO.sub.2 was introduced
as quartz, Fe.sub.2O.sub.3 as Fe.sub.2O.sub.3, and MgO as
Mg(OH).sub.2.(MgCO.sub.3).sub.4.(H.sub.2O).sub.5. Conventional
transmission .sup.57Fe Mossbauer spectroscopy measurements on
powdered samples were used to determine the iron redox state of the
untreated iron-containing glasses. A constant acceleration
spectrometer with a source of .sup.57Co in rhodium was used. The
spectrometer was calibrated using a foil of a--Fe at room
temperature. The ratio [Fe.sup.3+]/[Fe.sub.tot], where
[Fe.sub.tot]=[Fe.sup.2+]+[Fe.sup.3+], was found to be approximately
0.7 for both glasses. The T.sub.g values of 6 wtFe and 3 wtFe were
measured using differential scanning calorimetry (DSC), and found
to be 926 K and 921 K, respectively.
[0207] The obtained glasses were cut in cylinders and then ground
by a six-step procedure with SiC paper under ethanol, followed by
polishing with 1 .mu.m diamond suspension. Heat-treatments in the
H.sub.2/N.sub.2 (1/99) atmosphere were conducted at 1 atm in an
electrical furnace. The glass samples were inserted into the cold
furnace and the gas-flow was turned on. Heating and cooling of the
furnace were conducted at 10 K/min. Treatments in CO/CO.sub.2
(98/2) were conducted similarly, but the heating and cooling rate
was 5 K/min. The partial pressure of oxygen was kept at a known
value in the H.sub.2/N.sub.2 (1/99) atmosphere by using a
Fe.sub.3O.sub.4/Fe.sub.2O.sub.3 redox buffer. Fe.sub.2O.sub.3 and
Fe.sub.3O.sub.4 powders were mixed in the molar ratio 3:2 and
placed inside the furnace together with the samples. In the
CO/CO.sub.2 (98/2) atmosphere, the oxygen partial pressure was
controlled by the CO-00.sub.2--O.sub.2 equilibrium.
[0208] Fourier transform infrared (FT-IR) and
ultraviolet-visible-near infrared (UV-VIS-NIR) absorption spectra
were measured using doubly polished 0.2 mm thick glass slides with
Bruker Vertex 70 FT-IR and Analytik Jena UV-VIS-NIR Specord 200
spectrophotometers, respectively. From FT-IR spectra, the
permeation of H.sub.2 and CO into the glasses can be investigated
as incorporated OH and CO.sub.3 groups are detectable in IR
spectra. UV-VIS-NIR spectra were recorded to determine the change
in the iron redox state as a function of heat-treatment conditions.
The Fe.sup.2+ ion has a maximum absorption peak near 1050 nm but
the position and intensity of this peak varies with glass
composition. The absorption coefficients for our glasses were not
known and therefore the absorption coefficient of the Lambert-Beer
equation was calculated: A=c.epsilon.t, where A is absorbance, c
the concentration, E the absorption coefficient, and t the sample
thickness. By using the [Fe.sup.3+]/[Fe.sub.tot] ratio and the
total iron content, the concentration of ferrous iron was
calculated in the untreated 6 wtFe glass. Plotting the absorbance
near 1050 nm versus the sample thickness (0.12, 0.20, 0.40, and
0.80 mm) gave a linear relation (R.sup.2=0.997). From the slope of
this plot (c.epsilon.), the absorption coefficient was calculated
to be 3.90 L mol.sup.-1 mm.sup.-1.
[0209] To study the cationic diffusion processes, compositional
analysis of the glass surfaces was carried out using electron-gas
secondary neutral mass spectroscopy (SNMS) with an INA 3 (Leybold
AG) instrument. The analyzed area had a diameter of 5 mm and was
sputtered using Kr plasma with an energy of .about.500 eV. The time
dependence of the sputter profiles was converted into depth
dependence by measuring the depth of the crater at 10 different
locations on the same sample with a Tencor P1 profilometer.
[0210] Two properties of the heat-treated glasses were tested.
Vickers hardness was measured 25 times for each sample using a
Struers Duramin 5 microindentor at a load of 0.25 N and a hold time
at the maximum load of 5 seconds. The lengths of the indentation
diagonals were measured using an optical microscope (reflection
method). Chemical durability was tested by measuring leached
amounts of Na.sup.+ and Mg.sup.2+ ions after dissolution in 0.25 M
HCl and KOH solutions. The samples were immersed in plastic
containers with 20 cm.sup.3 test solution for each 1 cm.sup.2 of
the glass surface area. The containers were mounted on a
thermostatic shaking assembly at 90.degree. C. (agitated at 100
ppm) and after 12 h, the samples were removed from the solutions.
Atomic absorption spectroscopy (AAnalyst 100, Perkin Elmer) was
employed to measure the concentrations of Na.sup.+ and Mg.sup.2+ in
the test solutions.
[0211] Results and discussion. FIG. 16 shows UV-VIS-NIR spectra of
glasses heat-treated at T.sub.g for 16 h in H.sub.2/N.sub.2 (1/99)
or CO/CO.sub.2 (98/2), respectively. A maximum absorption peak is
seen near 1050 nm, which is attributed to the existence of the
Fe.sup.2+ ions. When the glass is heat-treated in H.sub.2/N.sub.2
(1/99) or CO/CO.sub.2 (98/2) for a given duration, the intensity of
the Fe.sup.2+ band increases, indicating that Fe.sup.3+ is reduced
to Fe.sup.2+. The change in Fe.sup.2+ concentration
(.DELTA.(Fe.sup.2+)) increases approximately linearly with the
square root of the heat-treatment duration (t.sub.a.sup.0.5),
implying that diffusion-limited kinetics occurs (see inset of FIG.
16). The effect of heat-treatment of the glasses was measured by IR
(not shown). Bands at 3550 and 2850 cm.sup.-1 are caused by O--H
stretching vibrations of weakly and strongly hydrogen-bonded OH
species, respectively. Bands near 1860 and 1630 cm.sup.-1 can be
assigned to combination modes and overtones of the silica glass
matrix. The bands positioned at 1510 and 1425 cm.sup.-1 are
assigned to vibrations of chemically dissolved carbonate species.
One of the carbonate oxygens is attached to a tetrahedral site via
a non-bridging oxygen (NBO). This complex is associated with
Ca.sup.2+. The following reactions account for the observed bands
as a result of treatments in H.sub.2/N.sub.2 (1/99) and CO/CO.sub.2
(98/2):
H.sub.2+2 NaFe.sup.3+O.sub.2+4 SiOSi.fwdarw.4
SiO(Fe.sup.2+).sub.0.5+2SiOH+2SiONa
CO+2
NaFe.sup.3+O.sub.2+SiOCa.sub.0.5+3SiOSi.fwdarw.4SiO(Fe.sup.2+).sub.-
0.5+SiCO.sub.3Ca.sub.0.5+2SiONa
[0212] In the notation, the formulas depict the bonding environment
of the oxygen anions. NaFe.sup.3+O.sub.2 represents a Fe.sup.3+,
which is tetrahedrally coordinated with oxygen and charge-balanced
by Na.sup.+. SiOSi corresponds to a bridging oxygen connecting two
silica tetrahedra. SiOH is a silica tetrahedron containing a
hydroxyl group. SiCO.sub.3Ca.sub.0.5 is a carbonate species
connected to a NBO and Ca.sup.2+. SiO(Fe.sup.2+).sub.0.5, SiONa,
and SiOCa.sub.0.5 represent that Fe.sup.2+ (octahedral
coordination), Na.sup.+, and Ca.sup.2+ are connected to a NBO,
respectively. In summary, the results show that both H.sub.2 and CO
are capable of permeating into the glass. Fe.sup.3+ is reduced to a
greater extent in H.sub.2/N.sub.2 (1/99) than in CO/CO.sub.2
(98/2), even though the CO partial pressure is much higher than
that of H.sub.2. This is explained by the faster permeation rate of
a H.sub.2 molecule due to its smaller size. Based on the covalent
radii of H, C (sp), and O, we have calculated the lengths of
H.sub.2 and CO molecules to be 1.2 and 2.7 .ANG., respectively.
[0213] The SNMS depth profile of the 6 wtFe glass heat-treated in
CO/CO.sub.2 (98/2) at its T.sub.g for 16 h is not shown. However, a
pronounced decrease of the concentration of Mg.sup.2+, Ca.sup.2+,
and Fe.sup.2+ towards the surface is observed (thickness: 300-350
nm). Na.sup.+ also diffuses towards the interior. Even though
alkali ions are normally found to be faster than earth alkaline
ions in glasses due to their lower charge, the diffusion depth of
Na.sup.+ is smaller than that of Mg.sup.2+, Ca.sup.2+, and
Fe.sup.2+, which is in agreement with the above studies. The inward
diffusion occurs to charge-balance the outward flux of electron
holes, and the charge might be most effectively transferred by the
divalent cations.
[0214] It should be noted that an enrichment of Na.sup.+ is
observed in the depth interval from 100 to 150 nm. This is ascribed
to the depletion of Mg.sup.2+, Ca.sup.2+, and Fe.sup.2+ ions in
this range because their depletion causes a relatively high
concentration of Na.sup.+ ions. It should also be noted that the
surface depletion of network-modifying cations is not due to the
polishing procedure for two reasons. First, the glass was ground
using SiC papers under ethanol and polished using a diamond paste,
i.e., no leaching of cations should occur. Second, a SNMS profile
of the untreated glass does not show any inward diffusion of
cations.
[0215] The SNMS profile of the glass treated in CO/CO.sub.2 (FIG.
17) indicates that the mechanism of Fe.sup.3+ reduction in
CO/CO.sub.2 (98/2) is the same as that in H.sub.2/N.sub.2 (1/99).
The internal reduction of Fe.sup.3+ generates electron holes
(h.sup. ). An outward flux of h.sup. occurs, which is driven by the
gradient in oxygen activity across the reaction zone. h.sup. are
filled by electrons released by ionic oxygen at the surface since
oxygen is released into the reducing atmosphere as CO.sub.2. The
outward flux of h.sup. is accompanied by inward flux of
network-modifying cations to maintain the charge-balance. Hence,
the inward cationic diffusion is driven by reduction of the high
valence to the low valence state of the polyvalent cation. To
explore whether or not the reaction is rate-limited by the
diffusion of divalent cations, diffusion coefficients for the
divalent cations should be calculated and compared to known values
of diffusion coefficients for divalent cations in similarly
polymerized glasses. The diffusion coefficient for a divalent
cation (D.sub.M.sub.2+) can be calculated by using the following
equation:
D M 2 + = .DELTA. .xi. 2 X M 2 + .DELTA. t ln ( p O 2 '' p O 2 ' )
, ##EQU00009##
where .DELTA..xi. is the thickness of the modifier layer,
X.sub.m.sub.2+ is the cation mole fraction of the divalent cation
M.sup.2+, .DELTA.t is the reaction time, p'.sub.O.sub.2 is the
partial pressure (i.e., activity) of oxygen at the free surface,
and p''.sub.O.sub.2 is partial pressure of oxygen at the internal
reaction front. p'.sub.O.sub.2 is fixed by the
CO--CO.sub.2--O.sub.2 equilibrium and is equal to 510.sup.-27 bar
at T.sub.g=653.degree. C. p''.sub.O.sub.2 depends on the initial
iron redox ratio and is calculated to be approximately 510.sup.-3
bar at T.sub.g=653.degree. C. Inserting these values into the above
equation gives a diffusion coefficient of Fe.sup.2+ cations in
CO/CO.sub.2 (98/2) of .about.110.sup.-18 m.sup.2/s. The value
agrees well with diffusion measurements for divalent,
network-modifying cations in glasses of similar polymerization.
This provides clear evidence for the mechanism. Hence, both CO
permeation and outward flux of electron holes contribute to the
reduction of Fe.sup.3+. In summary, the inward cationic diffusion
causes the creation of a silica-rich surface layer in the
iron-containing glasses. For the 3 wtFe glass treated in
CO/CO.sub.2 (98/2) at its T.sub.g for 16 h, the layer thickness is
.about.200 nm. This implies that when lowering the concentration of
Fe.sup.3+ ions, the extent of divalent ionic diffusion decreases,
and therefore, the layer becomes thinner. The layer is also created
when the 6 wtFe glass is heat-treated in H.sub.2/N.sub.2 (1/99).
However, in this case the thickness is .about.600 nm at T.sub.g for
16 h, which gives a value of D.sub.M.sub.2+ of .about.510.sup.-18
m.sup.2/s. This suggests that H.sub.2 be more effective in creating
the silica-rich surface than CO even though the oxidation potential
of CO is larger than that of H.sub.2 at 926 K (T.sub.g). The
difference in the layer thickness must then be due to the
difference in size of the two gaseous molecules. To neutralize the
electron holes at the surface, H.sub.2 and CO molecules must first
penetrate into the uppermost surface layer, subsequently be
dissolved in the structure, and simultaneously contact and reduce
the ferric ions in the glass structure. The penetration, and hence,
reduction process is easier when the molecule is small.
[0216] The hardness and chemical resistance of the untreated and
heat-treated samples are displayed in Table 10. From the structural
point of view, the earth alkaline and alkali cations disrupt the
continuous Si--O random network, and so introduce NBOs to the
glasses. Their removal from the surface clearly increases the
hardness and chemical resistance of the glasses. The increase is
most pronounced as a result of the H.sub.z-treatment as treatment
in this atmosphere creates the thickest silica-rich layer.
TABLE-US-00010 TABLE 10 Effect of the atmosphere of the
heat-treatment on the Vickers hardness (H.sub.v) and chemical
durability of the 6wtFe glasses. Property Untreated 98 vol % CO 1
vol % H.sub.2 H.sub.v (GPa) 8.9 .+-. 0.2 9.3 .+-. 0.2 9.9 .+-. 0.3
C(Na.sup.+).sub.acid (mg/L) 8.7 .+-. 0.3 5.1 .+-. 0.2 1.9 .+-. 0.1
C(Mg.sup.2+).sub.alkali (mg/L) 2.4 .+-. 0.3 1.6 .+-. 0.1 1.3 .+-.
0.1 The treated samples have all been heated at T.sub.g = 926 K for
16 h. Chemical durability of the glasses is expressed by the
leached amount of Na.sup.+ after 12 h in 0.25M HCl solution
(C(Na.sup.+).sub.acid) and Mg.sup.2+ after 12 h in 0.25M KOH
solution (C(Mg.sup.2+).sub.alkali).
Conclusions. A silica-rich surface layer can be created by
heat-treating an iron-bearing glass at its T.sub.g in both CO- and
H.sub.2-containing atmospheres. The layer is created due to the
inward diffusion of network-modifying cations. By calculating the
diffusion coefficient for the divalent cations, we have clarified
the mechanism of the inward diffusion. The glass surface becomes
structurally more polymerized due to the removal of
network-modifying cations from the surface. Consequently, the
hardness and chemical durability of the glasses are enhanced. In
addition, it is found that the extent of the inward diffusion is
larger as a result of the H.sub.2-treatment than of the
CO-treatment. This is attributed to the fact that H.sub.2 has a
smaller size than CO, so that the former more readily reduces the
ferric ions in the surface structure than the latter.
On-Going Experimental Work
[0217] For the on-going experimental work of the inventors, the
following points of consideration may or have been investigated in
further details by experimental and/or theoretical means and
methods: [0218] Further elucidation of different glass compositions
for achieving a silica nanolayer. This includes: [0219] Finding the
lowest limit of the concentration of the polyvalent ion in the
glass to enable the invention, in particular for practical
implementation. [0220] Creating a SiO.sub.2-rich surface nanolayer
in normal medical glasses (borosilicate glasses, wt %: 75-80
SiO.sub.2, 10-13 B.sub.2O.sub.3, 2-5 Al.sub.2O.sub.3, 4-7
Na.sub.2O, 0-2 CaO) or in window glasses by adding a polyvalent
element. [0221] Further elucidation of the impact on the
concentration of the H.sub.2 in the gas mixture affecting the
formation of the nanolayer. [0222] Dependent on glass composition,
H.sub.2 will be more or less soluble. Hence, different conc. of
H.sub.2 will result in inward diffusion (probably: when H.sub.2 is
more soluble, a lower H.sub.2 conc. is needed to obtain inward
diffusion). [0223] Further elucidaton of the physical and
mathematic model for describing the mechanism of the inward
diffusion in glasses. [0224] Characterizing the nanostructured
layer more precisely. [0225] TEM-images of the surface layers
(untreated, periclase layer and silica-rich layer) can give insight
into the structure of these layers. [0226] XPS: characterization of
the uppermost (few nm) layer [0227] Systematic CEMS (conversion
electron Mossbauer spectroscopy) investigations [0228] Finding the
optimum temperature and duration of heat-treatment for achieving
the nanolayer on different glass types. [0229] Further elucidation
of the created nanolayers impact on the properties of glasses.
[0230] hardness (by nanoindentation) [0231] chemical durability in
different solutions [0232] optical: antireflection, IR-absorption
(heat), refractive index, etc. [0233] T.sub.g of the surface layer
[0234] high temperature stability: heat modified surface in air at
elevated temperature and measure roughness before and after
heating. [0235] Analyze whether a silica-rich nanolayer can be
created on glass fibres that are used in composite materials.
[0236] It should be noted that embodiments and features described
in the context of one of the aspects of the present invention also
apply to the other aspects of the invention.
[0237] All patent and non-patent references cited in the present
application, are hereby incorporated by reference in their
entirety.
REFERENCES
[0238] Pind M. and Sorensen P. M. (2004) Effect of the redox state,
iron content and silica/alumina ratio on the crystallization
be-haviour of iron-bearing aluminosilicate glasses, Master thesis,
Aalborg University, Denmark. [0239] V. Rigato, G. Della Mea, Carlo
G. Pantano (1994) "Hydrogen profiles in the surface of reduced
lead-silicate glasses", Surface and Interface Analysis, Volume 21,
Issue 2, Pages 144-149. [0240] Deriano S., Rouxel T., Malherbe S.,
Rocherulle J., Duisit G., Jezequel G. (2004) "Mechanical strength
improve-ment of a soda-lime-silica glass by thermal treatment under
flowing gas", Journal of the European Ceramic Society, 24,
2803-2812. [0241] Gaillard F., Schmidt B., Mackwell S., and
McCammon C. (2003) "Rate of hydrogen-iron redox exchange in
silicate melts and glasses", Geochimica et Cosmochimica Acta, 67,
2427-2441. [0242] Gersten J. I. and Smith F. W. (2001) The Physics
and Chemistry of Materials, John Wiley & Sons, New York USA.
[0243] L. D. Pye, A. Montenero, and I. Joseph (2005) Properties of
Glass-Forming Melts, CRC Press. [0244] J. E. Shelby (2005)
Introduction to Glass Science and Technology, The Royal Society of
Chemistry, Cambridge, UK. [0245] Stanworth J. E. (1971) "Oxide
Glass Formation from the Melt", Journal of the American Ceramic
Society, 54, 61-63. [0246] Vogel W. (1994) Glass Chemistry,
Springer Verlag, Berlin, Germany. [0247] Zachariasen W. H. (1932)
"The Atomic Arrangement in Glass", Journal of the American Chemical
Society, 54, 3841-3851. [0248] Zotov N., Yanev Y., Epelbaum M., and
Konstantinov L. (1992) "Effect of water on the structure of
rhyolite glasses--X-ray diffraction and Raman spectroscopy
studies", Journal of Non-Crystalline Solids, 142, 234-246.
* * * * *